plant assemblage structure on barrier island
TRANSCRIPT
PLANT ASSEMBLAGE STRUCTURE ON BARRIER ISLAND
‘PIMPLE’ DUNES AT THE VIRGINIA COAST RESERVE
LONG-TERM ECOLOGICAL RESEARCH SITE
by
Brett A. McMillan M.S. May 1999, University of Florida
B.A. May 1996, Berea College
A Dissertation Submitted to the Faculty of Old Dominion University in Partial Fulfillment of the
Requirement for the Degree of
DOCTOR OF PHILOSOPHY
ECOLOGICAL SCIENCES
OLD DOMINION UNIVERSITY December 2007
Approved by:
________________________ Frank P. Day (Director)
________________________ Joseph Rule (Member)
________________________ G. Richard Whittecar (Member)
________________________ Donald R. Young (Member)
ABSTRACT
PLANT ASSEMBLAGE STRUCTURE ON BARRIER ISLAND ‘PIMPLE’ DUNES AT THE VIRGINIA COAST RESERVE
LONG-TERM ECOLOGICAL RESEARCH SITE
Brett A. McMillan Old Dominion University, 2007
Director: Dr. Frank P. Day
The habitats at the VCR LTER that were the focus of the current study are the
Hog Island and Parramore Island ‘pimples’, small, rounded dunes forming along main
dune ridges of the barrier islands. There are distinct plant assemblage zones found on
pimples, although most of these dunes are 10 – 20 m in diameter. Hypotheses of the study
were that fresh water availability was a main determinant of differences between
assemblages and that pimple size and location would influence diversity and assemblage
structure. Research goals were 1) to describe the plant assemblages on pimples, 2) to
relate edaphic and geomorphological factors to pimple assemblage diversity and
composition at different spatial scales, and 3) to compare assemblage – environment
interactions on pimples and main dune ridges. Accomplishing these goals entailed field
vegetation surveys of a representative sample of pimple and dune plant assemblages and
environmental monitoring. There were distinct assemblage types that segregated
themselves by habitat type: marsh, shrub thicket, and dry summit. Shrub assemblages
were less diverse than either marsh or summit habitats. There was no relationship
between pimple size and diversity or location. Differences in diversity and composition
among pimples were as great as differences among transects within pimples. Pimple
diversity and species composition were different from the main dunes. Fresh water
availability was important in differentiating differences, both among transects and among
species, but it was not the only important factor. Nutrients, such as boron, were also
important in describing variation among species. It is likely that interactions between
water and other factors (e.g. the accumulation of some mineral nutrients in the marsh
after they are leached from the dune summits) are the most important determinants to
species abundances. A secondary goal was to evaluate ordination techniques used in
pattern detection throughout the study. Canonical correspondence analysis and non-
metric multidimensional scaling performed best overall. CCA, which is a direct gradient
analysis, described groups of transects and species that largely matched my a priori
assumptions. Furthermore, it provided correlation data about species – environment
relationships that were equivalent to multiple regression. NMS, a distance-based,
indirect-gradient method, described high percentages of variation (> 80 %) in the first
two or three axes, but relationships between environment and species abundances had to
be inferred. Bray-Curtis ordination and especially principal components analysis did not
explain as much variation in the data.
iv
This thesis is dedicated to M.E.M., T.A.M., and R.D.B.
v
ACKNOWLEDGMENTS
Many thanks are due to Dr. Frank Day. He was a rarity among advisers in that he
actually advised, and he did so competently, compassionately, and consistently. The staff
at the VCR LTER made my work possible with boat trips and logistical support. My
committee, Drs. Joseph Rule, Rich Whittecar, and Don Young steered me in several good
directions. Drs. Tom Crist, Robert Colwell, and Nick Gotelli aided me in understanding
their statistical programs. Drs. Kate Lyons and Lisa Horth gave me great suggestions for
the direction of my thesis, many in the eleventh hour. Mike Mailand, Rachel Nodurft,
Christina Morgan, Stacey Galuffo, and Shannon Davis provided invaluable assistance,
memory, and sanity in both field and lab. Without Margot (Meg) Miller in the department
of Environmental Sciences at the University of Virginia, I would not have been able to
complete my nitrogen analysis. Dr. Rebecca D. Bray helped with species identifications,
oversaw the reposition of the plant voucher collection, and offered welcome advice or
swift kicks, whichever the situation demanded. My colleagues Alisha Pagel Brown,
Daniel Stover, Jay Bolin, Carmony Hartwig, and Deborah Hutchinson helped me solve
innumerable problems with my research and my life. Scott Derosier offered editorial
advice. Financial support was provided by subcontract 5-26173 through the University of
Virginia’s National Science Foundation LTER grant (NSF 0080381).
vi
TABLE OF CONTENTS
Page
LIST OF TABLES............................................................................................................ vii
LIST OF FIGURES ........................................................................................................... ix
Chapter
1. INTRODUCTION .............................................................................................1 BARRIER ISLAND ECOLOGY.................................................................1 THE VIRGINIA COAST RESERVE LONG-TERM ECOLOGICAL
RESEARCH SITE ...........................................................................8 THE PIMPLES OF THE VCR LTER .......................................................10 RESEARCH GOALS AND HYPOTHESES ............................................15
2. EXPERIMENTAL DESIGN AND METHODOLOGY .................................17 INTRODUCTION .....................................................................................17 PILOT STUDY..........................................................................................17 FINAL SAMPLING CONFIGURATION.................................................19 SPECIES COLLECTION..........................................................................22
3. PLANT ASSEMBLAGE STRUCTURE AND DIVERSITY PATTERNS....23 INTRODUCTION .....................................................................................23 MATERIALS AND METHODS...............................................................30 RESULTS ..................................................................................................38 DISCUSSION ............................................................................................55 CONCLUSIONS........................................................................................60
4. ENVIRONMENTAL INFLUENCES ON SPECIES DISTRIBUTION .........63 INTRODUCTION .....................................................................................63 MATERIALS AND METHODS...............................................................64 RESULTS ..................................................................................................70 DISCUSSION ..........................................................................................113 CONCLUSIONS......................................................................................122
5. CONCLUSIONS............................................................................................123 WATER AND OTHER DRIVERS .........................................................123 SPATIAL RELATIONSHIPS .................................................................125 STATISTICS ...........................................................................................125 CONCLUSION........................................................................................125
APPENDIX......................................................................................................................128
LITERATURE CITED ....................................................................................................133
VITA ................................................................................................................................159
vii
LIST OF TABLES
Table Page
1. Calculation of Shannon and Simpson’s diversity indices (Hill 1973a, Crist and Veech 2006). ..........................................................................................................31
2. Species abundance measures used in calculating species importance values (Curtis and McIntosh 1951, McIntosh 1962, Bannister 1966, Will-Wolf 1980)...32
3. Formulas for species richness estimators that approximate rarefaction curves. ..........34
4. Three measures of β-diversity (Whittaker 1960, Wilson and Shmida 1984, Shmida and Wilson 1985, Magurran 1988)...........................................................37
5. Terms and formulas used in additive diversity partitioning (Crist et al. 2003, Olszewski 2004, Crist and Veech 2006)................................................................39
6. Relative dominance and relative frequency for pimple species...................................44
7. Comparison of species importance value (modified), frequency, and dominance scores for species found in the main dunes of Hog Island and the summits of pimples...................................................................................................................46
8. Explanatory value of first three axes of Bray-Curtis ordination and non-metric multidimensional scaling and the ten most important species in each based on correlation (Pearson’s r).........................................................................................52
9. Diversity partitioning within and among pimples and within and among the main dunes as calculated with PARTITION. .................................................................55
10. Pearson’s correlation coefficients for environmental factors used in the study. .........73
11. Pearson’s r correlation coefficients of environmental variables for the first three axes of a CCA of main dunes and pimples and percentage of variation explained. ...............................................................................................................75
12. Pearson’s r correlation coefficients for the three axes of a NMS solution of main dune and pimple transects versus environmental factors.......................................85
13. Pearson’s r correlation coefficients for three axes of a CCA solution describing the relationship between pimple transect assemblages and environmental factors.....................................................................................................................88
14. Pearson’s r correlation coefficients for two axes in a non-metric multidimensional scaling ordination of pimple transects based on environmental factors. ...............93
15. Pearson’s r correlation coefficients between environmental factors and three axes of a canonical correspondence analysis ordination of dune and pimple species. ..98
16. Pearson’s r correlation coefficients between environmental factors and three axes of a non-metric multidimensional scaling ordination of pimple and dune species. .................................................................................................................103
viii
17. Pearson’s r correlation coefficients between environmental factors and three axes of a canonical correspondence analysis ordination of pimple species.................106
18. Pearson’s r correlation coefficients between environmental factors and three axes of a non-metric multidimensional scaling ordination of pimple species. ............111
19. Correlation of environmental factors with the first three axes of a Bray-Curtis ordination of main dunes and pimples (Pearson’s r). ..........................................128
20. Pearson’s r correlation coefficients for the first three axes of a principal components analysis of main dune and pimple transects versus environmental factors...................................................................................................................128
21. Pearson’s r correlation coefficients for three axes of a Bray-Curtis ordination of pimple transects based on environmental factors. ...............................................129
22. Pearson’s r correlation coefficients for three axes of a principal components analysis ordination of pimple transects based on environmental factors.............129
23. Pearson’s r correlation coefficients between environmental factors and three axes of a Bray-Curtis ordination of pimple and dune species......................................130
24. Pearson’s r correlation coefficients between environmental factors and three axes of a principal components analysis of pimple and dune species. ........................130
25. Pearson’s r correlation coefficients between environmental factors and three axes of a Bray-Curtis ordination of pimple species. ....................................................131
26. Pearson’s r correlation coefficients between environmental factors and three axes of a principal components analysis ordination of pimple species........................132
ix
LIST OF FIGURES Figure Page
1. Virginia’s Eastern Shore and the VCR – LTER sites. .................................................11
2. Profile of a small pimple, not to scale; diameter varies from 10 to 20 m....................13
3. Sampling method. ........................................................................................................19
4. Map and aerial photograph of pimples and main dunes sampled on Hog Island. .......21
5. Per-pimple and per-transect mean species richness by year with grand total mean. ...40
6. Maps showing pimple sizes, richness, Shannon diversity, and Simpson’s diversity..................................................................................................................40
7. Relationships of pimple size and location to richness. ................................................41
8. Yearly mean transect-level species richness in each habitat type................................42
9. Yearly mean transect-level Shannon diversity in each habitat type. ...........................42
10. Yearly mean transect-level Simpson’s diversity index in each habitat type . .............42
11. Rarefaction curves of species encountered based on number of transects sampled. ...43
12. Histogram of pimple species frequency.......................................................................47
13. Rarefaction curve showing partitions of diversity, observed species richness (Sobs) with confidence intervals, and two richness estimators: ICE and Chao 2. ............48
14. Three axes of a Bray-Curtis ordination of transects based on species abundances. ....49
15. Non-metric multidimensional scaling of transects based on species abundances. ......50
16. Principal components analysis for transects based on species abundances. ................51
17. Comparison of three measures of beta diversity between transects in pimples by year (Magurran 1988). ...........................................................................................53
18. Maps of three measures of beta diversity between transects in pimples. ....................54
19. Diversity partitioning within pimples as measured by three diversity indices. ...........54
20. Content of several nutrients, base saturations, organic matter content, and cation exchange capacity for marsh, shrub, and xeric habitat soils on pimples and main dunes. ............................................................................................................71
21. Ammonium, nitrate–nitrite, phosphorus, and zinc; organic horizon thickness; salinity; % slope, and water table for three different habitats on pimples and main dunes. ............................................................................................................72
22. Canonical correspondence analysis ordination of main dune and pimple transects....75
23. Overlay of water table position on the CCA ordination of pimple and main dune transects..................................................................................................................76
24. Overlay of soil boron concentration on the CCA ordination of pimple and main dune transects.........................................................................................................77
x
25. Overlay of soil zinc concentration n on the CCA ordination of pimple and main dune transects.........................................................................................................78
26. Overlay of soil magnesium concentration on the CCA ordination of pimple and main dune transects................................................................................................79
27. Overlay of soil iron concentration on the CCA ordination of pimple and main dune transects.........................................................................................................80
28. Overlay of soil phosphorus concentration on the CCA ordination of pimple and main dune transects................................................................................................81
29. Overlay of soil nitrate – nitrite concentration on the CCA ordination of pimple and main dune transects. ........................................................................................82
30. Overlay of soil cation-exchange capacity on the CCA ordination of pimple and main dune transects................................................................................................83
31. Overlay of soil organic matter concentration on the CCA ordination of pimple and main dune transects................................................................................................84
32. Non-metric multidimensional scaling ordination of pimple and dune transects based on environmental factors. ............................................................................86
33. Overlay of potassium concentration on pimple and dune transects in a non-metric multidimensional scaling ordination......................................................................86
34. Overlay of water table height on pimple and main dune transects in a non-metric multidimensional scaling ordination......................................................................87
35. Overlay of soil ammonium concentration on pimple and dune transects in a non-metric multidimensional scaling ordination...........................................................87
36. Canonical correspondence analysis of pimple transects with all recorded environmental variables. ........................................................................................89
37. Overlay of elevation above marsh on a CCA of pimple transects. ..............................89
38. Overlay of water table position above the mean minimum level on a CCA of pimple transects. ....................................................................................................90
39. Overlay of soil magnesium concentration on a CCA of pimple transects. ..................90
40. Overlay of cation-exchange capacity on a CCA of pimple transects. .........................91
41. Overlay of soil phosphorus concentration on a CCA of pimple transects...................91
42. Overlay of soil iron concentration on a CCA of pimple transects. ..............................92
43. Non-metric multidimensional scaling ordination of pimple transects. ........................93
44. Overlay of water table position on a NMS ordination of pimple transects. ..............94
45. Graphic overlay of elevation on pimple transects in an NMS ordination....................94
46. Overlay of salinity on pimple transects in an NMS ordination. ..................................95
47. Overlay of ammonium concentrations on pimple transects in an NMS ordination.....95
xi
48. Overlay of magnesium concentration on pimple transects in an NMS ordination. .....96
49. CCA of main dune and pimple species based on mean environmental variables. ......98
50. Overlay of soil boron on a CCA ordination of main dune and pimple species. ..........99
51. Overlay of soil zinc on a CCA ordination of dune and pimple species.......................99
52. Overlay of soil magnesium on a CCA ordination of dune and pimple species. ........100
53. Overlay of soil iron on a CCA ordination of dune and pimple species. ....................100
54. Overlay of soil organic matter on a CCA ordination of dune and pimple species. ...101
55. Overlay of soil ammonium on a CCA ordination of dune and pimple species. ........101
56. Overlay of mean position of water table above the average minimum on a CCA ordination of dune and pimple species.................................................................102
57. Overlay of NOx concentration on pimple and dune species in NMS ordination. ......103
58. Overlay of potassium base saturation on pimple and dune species in an NMS ordination. ............................................................................................................104
59. Overlay of magnesium concentration on dune and pimple species in an NMS ordination. ............................................................................................................104
60. Canonical correspondence analysis of pimple species based on environmental factors...................................................................................................................107
61. Overlay of mean soil potassium on pimple species in a CCA ordination. ................108
62. Overlay of soil ammonium on pimple species in a CCA ordination. ........................109
63. Overlay of mean soil iron concentrations on pimple species in a CCA ordination...110
64. Two-axis non-metric multidimensional scaling ordination of pimple species based on environmental variables. .................................................................................111
65. Overlay of potassium concentration on pimples in an NMS ordination....................112
66. Overlay of cation-exchange capacity on pimples in an NMS ordination. .................112
1
CHAPTER 1
INTRODUCTION
BARRIER ISLAND ECOLOGY
The factors influencing plant assemblages on dunes and barrier islands were
among the first subjects of interest to ecologists, one of the most notable being Cowles
(1899), who first described the process of floristic succession based on observations of
forest structure on the dunes of Lake Michigan. Because they are created by winds, tides,
and currents, barrier islands are geomorphologically and biologically dynamic. Storms,
for example, can create new dune ridges or wipe out old ones overnight (Dolan and
Hayden 1981, Dolan et al. 1988, Kochel and Wampfler 1989). The interaction and
fluctuation of geomorphology, storms, hydrology, topography, and nutrient availability
on barrier islands make plant assemblages change rapidly relative to most mainland
systems.
Geomorphology and storms
Topography on barrier islands typically includes a conspicuously parallel
sequence of dune ridges. These form as storms, tides, and wind deposit sand and create
new foredunes, which are subsequently stabilized by colonizing plants (Cowles 1899,
Godfrey et al. 1979, Roman and Nordstrom 1988, Ehrenfeld 1990, Hayden et al. 1995,
Anthonsen et al. 1996, Bate and Ferguson 1996, Rust and Illenberger 1996, Fearnehough
et al. 1998, Lichter 1998). The age of these dune ridges can be estimated through
radiodating; cartographic, historical, and meteorological records; and remotely-sensed
data, such as aerial survey photographs. Researchers use this evidence to establish a
This dissertation follows the format of Ecology.
2
‘chronosequence’ of dune ages across islands (Ehrenfeld 1990, Hayden et al. 1991,
Hayden et al. 1995, Day et al. 2001).
The effects of both salinity and flooding make storms a major force in shaping
barrier island plant communities. Salt spray can travel across entire islands during severe
storms, impacting interior species with low salinity tolerance (Ehrenfeld 1990). Saline
flooding from storm surges has an even greater impact on salt-intolerant species
(Ehrenfeld 1990, Young et al. 1994, Young et al. 1995a, Young et al. 1995b, Hester et al.
1996, Houle 1997). Storm surges and wind can either bury or wash away entire plant
assemblages as sand is deposited or eroded; dune plant species respond with adaptations
including increased growth rates, enhanced CO2 uptake, enhanced germination, and
varied biomass allocations between above and belowground tissue, depending on survival
strategy (Weller 1989, Yuan et al. 1993, Erickson 1994, Young et al. 1994, Erickson and
Young 1995, Brown 1997, Perumal and Maun 1999). The relative importance of salinity
tolerance and burial tolerance to the structure of plant assemblages has been studied, but
results are equivocal and often species-specific (Schroeder et al. 1979, Young et al.
1995a, Young et al. 1995b, Bate and Ferguson 1996, Rust and Illenberger 1996, Dilustro
and Day 1997, Ehrenfeld 1997).
Importance of land and water surfaces
On barrier islands, seemingly minute elevation changes can have a dramatic effect
on water availability and water quality, both of which directly influence biological
communities (Hayden et al. 1995, Zou et al. 1995, Lammerts et al. 2001, Muñoz-Reinoso
2001). Several processes, including evapotranspiration, astronomical and wind-driven
tides, rainfall and drought, influence the amount of surface and ground water available on
3
barrier islands and the salinity of that water. This is a major, if not the most important,
determinant of vegetation assemblage structure (Ehrenfeld 1990, Young et al. 1994,
Hayden et al. 1995, Tolliver et al. 1997).
Underground freshwater percolating into the soil from rain forms a lens-shaped
zone of saturation that floats on denser saline water derived from overwash or lateral
infiltration from the ocean. Theoretically, such as lens can form on a sand island of any
size, whether a pimple mound or a coastal barrier. It develops because of a combination
of rainwater infiltrating across the island and groundwater seepage discharging into
bodies of water around the island edge. Thus, a sand dune by itself may not affect the
elevation of the water table beneath it, but rather a dune (or pimple mound) flanked by
permanent water bodies is likely to generate a relatively high elevation on the water table
surface (Whittecar and Emry 1992).
The depressions between successive dune ridges, ‘slacks’ or ‘swales’, are prone to
flooding and often contain marsh communities. A distinct boundary between plant
species assemblages often exists at the junction of dunes and swales (Jones and
Etherington 1971, Godfrey et al. 1979). This is the result of differing abilities to
withstand the effects of flooding, primarily anoxia and mineral poisoning (e.g., from
Fe++, Mn++, or sulfides) related to chemically reduced conditions (Jones and Etherington
1971, Jones 1972a, 1972b, 1975a, 1975b, Studer-Ehrensberger et al. 1993). Salinity of
water in a swale marsh varies with elevation and exposure to tides and storm overwash.
Assemblages within and bordering the marsh are structured as a consequence of resident
species’ relative tolerances to salinity and desiccation (Godfrey et al. 1979, Young et al.
1994, Hayden et al. 1995, Young et al. 1995a).
4
Nutrient availability
In addition to periodic disturbance and changing microtopography, nutrient
availability influences community structure and function on barrier islands (Ehrenfeld
1990, Young et al. 1992, Verhoeven et al. 1996). In island ecosystems, patchy
availability of nutrients, especially nitrogen and phosphorus, influences species
composition and above and belowground biomass. Nutrient cycling is consequently tight,
with most nutrients sequestered in biomass (Day 1996, Stevenson and Day 1996,
Verhoeven et al. 1996, Dilustro and Day 1997).
The well-drained, sandy soils on coastal dunes have high leaching rates of nitrate
from the upper layers of the soil (Kachi and Hirose 1983, Sande and Young. 1992,
Verhoeven et al. 1996, Wijnholds and Young 2000). Cation-exchange capacity and,
hence, availability of most mineral nutrients are limited due to the low organic matter and
clay content of the soil (Brady and Weil 2002). Stressful conditions in dune soils, such as
low fresh water availability, may also inhibit nitrification and nitrogen mineralization
(Kachi and Hirose 1983).
In contrast to dunes, nutrient availability in swales is limited by an overabundance
of water. Flooding may directly limit the availability of some nutrients such as potassium
through leaching or dilution (Jones and Etherington 1971, Jones 1975b). By changing the
chemical species of nutrients to less bioavailable forms, introducing toxic forms of
elements such as reduced iron or aluminum, or altering the pH of the environment, anoxic
and reduced conditions associated with flooding may inhibit uptake of nutrients, such as
phosphorus and nitrogen, by some plant species (Jones 1975b).
5
Biotic interactions
Biotic factors also influence the distribution of nutrients in coastal ecosystems. In
swales, decompositional release of nutrients is inhibited by anoxia during flooding, and
nutrients are released in pulses whenever the soil dries out (Kushlan 1990, Conn and Day
1997). Plant–nitrogen-fixer symbioses, especially associations between shrubs of the
genus Morella (Myrica) and the nitrogen-fixing actinomycete Frankia, alter the nitrogen
content of the soil (Young et al. 1992, Young et al. 1994, Smith et al. 1995, Tolliver et al.
1997, Adler et al. 1998, Wijnholds and Young 2000). Besides adding nitrogen, leaf litter
from Morella increases soil nutrients directly through decomposition and indirectly
through increasing the cation exchange capacity of the soil (Young 1992, Smith et al.
1995, Adler et al. 1998).
There is evidence that vesicular-arbuscular mycorrhizae are important for the
success of dune colonization and stabilization by grasses (Koske and Polson 1984).
Moreover, mycorrhizal associations have been found important to the success and
abundance of Morella cerifera and other dune species; this implies that mycorrhizae may
play an important role in dune persistence as well (Semones and Young 1995, Field 1999,
Perumal and Maun 1999). Soil moisture and nutrient availability affect mycorrhizal
colonization in the dunes. These abiotic conditions may indirectly affect plant species
abundances through facilitation of symbiotic relationships as well as directly through
supplying nutritional requirements (Koske and Polson 1984, Al-Agely and Reeves 1995).
Community structure, succession, and state change
The relation of plant assemblage composition to age along dune chronosequences
in coastal and barrier island systems has been a major line of evidence for the existence
6
of ecological succession patterns (Cowles 1899, Levy 1990, Hayden et al. 1991, Avis and
Lubke 1996, Kerley et al. 1996, McLachlan et al. 1996, Crawford and Young 1998a,
Crawford and Young 1998b, Huggett 1998, Zonneveld 1999). Furthermore, much of the
largely semantic argument surrounding the concept of plant succession and the climax
community has involved studies of dune chronosequences (Clements 1936, Whittaker
1953, Olson 1958, Lichter 1998).
The serial succession of species replacements occurs rapidly on barrier islands
because species there are adapted for rapid colonization in a dynamic physical
environment (Ehrenfeld 1990, Levy 1990, Hayden et al. 1991, Hayden et al. 1995).
Succession in the coastal marshes is determined by a complex interaction of abiotic
stressors with interspecific competition and facilitation (Shumway and Bertness 1994,
Brewer and Bertness 1996, Bertness and Leonard 1997, Hacker and Bertness 1999). For
example, hypersalinity in newly created gaps in the marsh may favor salt-tolerant
pioneering species, e.g. Distichlis spicata (Bertness and Ellison 1987, Bertness 1991a,
Brewer and Bertness 1996, Brewer et al. 1998). Those species in turn ameliorate soil
salinity and facilitate establishment of species that are otherwise superior competitors,
e.g., Spartina patens (Bertness 1991b).
Establishment of a woody plant community on coastal dunes can happen fast
enough (sometimes <10 yr) to allow researchers to actually observe successional
processes rather than infer them (Ehrenfeld 1990, Johnson and Barbour 1990, Lichter
1998). In at least one North American coastal shrub species, Iva frutescens, there is an
apparent positive feedback effect of adult plants creating better habitats for recruiting
seedlings (Bertness and Yeh 1994, Hacker and Bertness 1995, Harley and Bertness
7
1996). Establishment of the climax-community maritime forest is not inevitable,
however, and herb or shrub-dominated assemblages may persist indefinitely on some
areas in response to disturbance regimes or other environmental conditions.
Frequent physical disturbances, such as storm overwash, as well as more gradual
processes, like accretion and sedimentation or sea level rise, bring about ecological state
changes in coastal areas like Virginia’s barrier islands that not only are often more rapid
than succession, but also can change its outcome (Hayden and Hayden 1994). For
example, a severe storm overwash event on an island could, in a relatively short period of
time, rearrange geomorphological features; alter nutrient cycling; change soil salinity and
microbial communities; and rapidly replace a shrub-dominated community with a grass-
dominated one.
Interspecific interactions, positive or negative, are also very important to plant
succession and the composition of plant assemblages; this has be most closely studied in
salt marsh systems(Bertness and Ellison 1987, Bertness 1991b, 1991a, Pennings and
Callaway 1992, Bertness and Hacker 1994, Bertness and Yeh 1994, Shumway and
Bertness 1994, Hacker and Bertness 1995, Shumway 1995, Brewer and Bertness 1996,
Brewer et al. 1998, Hacker and Bertness 1999, Costa et al. 2003). Research is beginning
to indicate that the coastal barrier landscape is a stress ‘mosaic’. Nearly every area on a
barrier island has a particular challenge: inundation and anoxia, high salt concentrations,
low nutrient concentrations, droughty soils, etc. Although many species (an example
from North America being Spartina alterniflora or S. patens) have a fairly broad
environmental tolerance, their range is restricted by resource competition with other
species (Bertness 1991b). Conversely, the distributions of other species (e.g., Iva
8
frutescens) seems to be contingent on positive interspecific interactions, such as soil
amelioration (Bertness and Hacker 1994, Bertness and Yeh 1994, Hacker and Bertness
1995).
THE VIRGINIA COAST RESERVE LONG-TERM ECOLOGICAL RESEARCH SITE
Site description
I undertook this research on the barrier islands at the Virginia Coast Reserve
Long-Term Ecological Research site, a multi-disciplinary, multi-institution ecological
research site consisting of coastal areas off the Eastern Shore of Virginia (Callahan 1984,
Franklin et al. 1990, Hayden et al. 1991, Olson et al. 1999). The Eastern Shore of
Virginia forms the southeastern edge of the Delmarva Peninsula, which is the
northeastern border of the Chesapeake Bay (Fig. 1). This region is situated on the coastal
plain along the trailing edge of the North American Plate (Hayden et al. 1991). The
seaside of the Delmarva Peninsula consists of a contiguous system of shallow bays
(lagoons), tidal flats, oyster shoals, inlets, salt marshes, and barrier islands (Dueser 1990,
Norcross and Hata 1990).
Virginia’s barrier islands represent about 8% of North America’s shoreline
(Hayden et al. 1991) and are the last remaining undeveloped stretch of coastline on the
mid-Atlantic seaboard (Badger and Kellam 1989, McCaffrey and Dueser 1990b). Located
3–20 km offshore, the islands are 2–14 km long, 1–2 km wide, and 1–9 m in elevation
above mean sea level (Hayden et al. 1991). They are centered on latitude 37.50º north
and longitude 75.66º west (McCaffrey and Dueser 1990a, 1990b). The tidal range is
approximately 1 m, and average seaside wave heights range from 0.5–1.0 m (Hayden et
al. 1991).
9
Research directives
A primary hypothesis of the VCR LTER project is that ecosystem, landscape, and land-
use patterns within terrestrial–marine watersheds are controlled by the vertical positions
of the land, the sea, and the freshwater table surfaces. Sub-hypotheses pertinent to the
barrier islands are 1) that dunes on the barrier islands are younger with proximity to the
ocean, and successional processes along the dune chronosequence are responsible for
biogeochemical variation across the landscape, 2) depth to the freshwater table and
magnitude of storm disturbance determine species composition and successional
processes, and 3) above and belowground productivity and decomposition rates are
functions of depth to the freshwater table and nitrogen availability. Studies conducted
there have, for example, focused on the roles of species life-history; probability of
general disturbance; storm overwash effects, which include sand burial and flooding; and
salinity tolerance from ground water or salt-spray (Schroeder et al. 1979, Fahrig 1990,
Levy 1990, Hayden et al. 1995, Crawford and Young 1998a).
Barrier island plant species
Although species composition varies among the islands, habitats and associated plant
assemblages can be generalized for the islands. Characteristically, there are salt marshes
on the lagoon side of the islands dominated by two growth forms of the halophytic grass
Spartina alterniflora Loisel (Godfrey et al. 1979, McCaffrey and Dueser 1990a,
McCaffrey and Dueser 1990b). Along lower dune slopes as well as the older swales,
there are upland shrub thickets dominated by Morella cerifera (L.) Small, Myricaceae,
along with M. pennsylvanica Loisel, Baccharis halimifolia L., and Iva frutescens L., both
Asteraceae (Young et al. 1994, Young et al. 1995b, Crawford and Young 1998b). Tops of
10
dune ridges support various drought-tolerant grasses, such as Spartina patens (Aiton)
Muhl., Ammophila breviligulata Fernald, Aristida tuberculosa Nuttall, or Schizachyrium
scoparium (Michx.) Nash, and a few drought-tolerant forbs. Swale marshes
predominantly contain the grasses Spartina patens and Distichlis spicata (L.) Greene;
other graminoids, Typha spp., and dicots, such as Hibiscus moscheutos L. and Phyla
(Lippia) lanceolata (Michx.) Greene, constitute a much smaller proportion of vegetation
cover (Godfrey et al. 1979, pers. obs.). On dunes of the largest islands, there are forests
composed mostly of loblolly pine, Pinus taeda L., with some dicot trees, e.g., Persea
borbonia (L.) Sprengel, Lauraceae (McCaffrey and Dueser 1990a, pers. obs.). Red cedar,
Juniperus virginiana L., Cupressaceae, is occasionally present both as a pioneer tree
species on developing dunes as well as a secondary species on older dunes (Young et al.
1994, Martin and Young 1997).
THE PIMPLES OF THE VCR LTER
The habitats at the VCR LTER that were the focus of the current study are the
‘pimples’ of Hog and Parramore Islands. Pimples are dune landforms or mounds that
superficially resemble other vegetation or tree islands found in other ecosystems.
Vegetation islands and pimples are relatively small assemblages of woody or otherwise
larger vegetation within a matrix of herbaceous species. Examples of tree islands can be
found in alpine meadows and tundra, midland prairies, and wetlands like the Florida
Everglades and Okefenokee Swamp (Rich 1934, Loveless 1959, Burbanck and Phillips
1983, Benedict 1984, Glasser 1985, Pauker and Seastedt 1996, Troxler Gann et al. 2005)
11
FIG. 1. Virginia’s Eastern Shore and the VCR – LTER sites.
.
12
Nevertheless, most of these community patches originate from circumstances
unique to their ecosystems. For example, tree islands in the Everglades arise from the
filling of karst solution holes with peat, whereas tree islands in the Okefenokee establish
on floating peat mats (Glasser 1985, Troxler Gann et al. 2005). Although there are
vegetation islands in marshes, pimples on the barrier islands of Virginia are underlain by
sand dunes, not peat, and so could not have arisen in either of those ways.
At the VCR, researchers first recognized and described pimple mounds on
Parramore Island, but pimples occur on several other islands, including Hog Island,
which is the most thoroughly studied of the chain (Rich 1934, Melton 1935, Dietz 1945,
Cross 1964). As opposed to more typical crescent-shaped or parabolic dunes, pimples are
circular to slightly ovate and flat-topped (Cross 1964, Anthonsen et al. 1996). Pimples are
typically < 2 m taller than the elevation of the surrounding marsh; diameters range from
10 to 100 m (Fig. 2; pers. obs.). Geologists and ecologists have speculated that the
pimples formed through various combinations and interactions of eolian deposition,
vegetation stabilization, and overwash erosion. Nevertheless, there is no conclusive
evidence of their actual origin (Rich 1934, Melton 1935, Dietz 1945, Cross 1964, Hayden
et al. 1995).
Their elevation above mean high water level allows the accumulation of an
underground freshwater lens floating on the saline groundwater (Hayden et al. 1995).
This freshwater lens supports upland plant species on the shoulders of the pimple, but
most pimples are too tall to support any but the most drought-tolerant species on their
summits (Hayden et al. 1995).
13
FIG. 2. Profile of a small pimple, not to scale; diameter varies from 10 to 20 m.
In aerial photographs of Parramore and Hog Island, pimples on both islands are
concentrated on the eastern side of swales, which makes them appear associated with the
younger, seaward dune ridge of the two ridges surrounding the swales. The largest
concentration of pimples on Parramore is in a swale whose foredune was washed out by
the Halloween or ‘Perfect’ Storm of 1991 (Young et al. 1995a). The lack of a foredune
exposes the Parramore pimples to frequent overwash, which creates a brackish to saline
marsh surrounding them (Hayden et al. 1995, Young et al. 1995a). Hog Island pimples,
however, are found in interior, freshwater swales that are largely protected from
overwash (Fahrig et al. 1993, Hayden et al. 1995, Young et al. 1995a). Parramore Island
pimples appear to have a greater mean diameter than Hog Island pimples and seem to
support richer and more diverse species assemblages. Some pimples on Parramore Island
have a concentric pattern of alternating shrub and grass zones, which Hayden et al.
(1995) hypothesized to reflect historical patterns of drought and rain events. No such
patterning is readily observable on Hog Island pimples. Tree species richness appears
greater on Parramore Island pimples than on Hog Island pimples.
xeric graminoids
shrubs
marsh graminoids
~1m
freshwater table
14
Biological processes are probably more important than geological ones in the
formation of pimples. There are what appear to be nascent pimples behind the youngest
dunes on Hog Island. These have younger shrubs or only beach grasses on them but are
similar in shape, size, and position (relative the main dune) as the interior pimples. This
suggests that they may be dune fragments whose coalescence has been blocked by the
deposition of newer dune. They persist because of stabilization by colonizing plants. The
composition of colonizing plant species has a major role determining the ‘mature’
morphology of barrier island dunes (Godfrey et al. 1979, Ehrenfeld 1990, Levy 1990,
Fahrig et al. 1993, Bailey et al. 1998). Because the VCR marks either the southern or
northern boundary of several North American coastal plain plant species, it is floristically
unique (Small 1933, Radford et al. 1968). This assemblage of plant (and perhaps
mycorrhizal) species may mean that pimples are unique, endemic geologic phenomenon
Since they are rare geomorphologic features, the pimples of Virginia’s barrier
islands have interested geologists for years (Rich 1934, Melton 1935, Dietz 1945, Cross
1964). Biologists have studied them less. The only major attempt to study the ecology of
pimples was when Hayden et al. (1995) initiated a groundwater displacement experiment
to evaluate the role of hydrology in the community ecology of the Parramore pimples, but
they were forced to abandon the project due to technical and logistical difficulties. It is
important to study the plant assemblages on pimples and the environmental factors
influencing them, not only because of the uniqueness of the pimples, but also to add to
the understanding of upland plant ecology of the islands in general, since terrestrial
systems in the VCR have received less attention than others. It is also unclear what
differences in assemblage structure and dynamics may exist between pimples and the
15
main dune ridges of the islands. Pimples appear to be semi-isolated dune ‘islands’ within
the islands, and this research will help to elucidate their similarities with the larger dunes
so that they may be used as research units or natural mesocosms for researching upland
ecology on barrier islands. Just as studying tree-fall gaps elucidates the dynamics of
forest succession and regeneration, studying pimples can illustrate the ecology of the
establishment and dynamics of the island system (Crawford and Young 1998a).
RESEARCH GOALS AND HYPOTHESES
My main hypotheses for this study were 1) that freshwater availability is the
primary factor determining assemblage structure on pimples and 2) that pimple size
influences the diversity of species assemblages. Research goals for testing my hypotheses
were 1) to describe the plant assemblages associated with pimples on Hog Island and
Parramore Island, 2) to relate the edaphic and geomorphological characteristics of the
pimples to the species composition of plant assemblages on them at island-level, pimple-
level, and sub-pimple-level scales, and 3) to compare assemblage – environment
interactions on pimples and main dune ridges. Accomplishing these goals entailed field
vegetation surveys of a representative sample of pimple and dune plant assemblages and
environmental monitoring.
Field measurements of assemblage structure were coupled with observations of
abiotic conditions, such as total C & N, pH, salinity, elevation, freshwater table depth and
fluctuation, and probability of storm overwash. Multivariate statistical techniques were
used to describe the relative importance of these factors to assemblage characteristics:
species diversity and richness, and the presence of certain species indicative of the
16
various assemblages. Studying pimple community dynamics on more than one island was
intended to increase the generality of inferences drawn from the study.
Based on the hypotheses of the VCR LTER project, a primary assumption of this
study was that the depth to freshwater and elevation would be the most important factors
determining plant assemblage structure on the pimples. It was more difficult to predict
other environmental factors contributing to variation in assemblage structure. Available
soil nitrogen would surely influence density, if not composition, of species on pimples.
Another prediction was that presence and density of the nitrogen-fixing shrub Morella
cerifera would affect species composition through modifying soil nitrogen content (Smith
et al. 1995). Between-island differences in species composition would be influenced by
water quality in the surrounding marsh, i.e., water conditions in the marsh would affect
pore water salinity and redox potential on pimples. Between-island differences could also
be explained by rate of disturbance, but estimates of storm overwash potential have only
been calculated for Hog Island (Kochel and Wampfler 1989, Fahrig et al. 1993, Hayden
et al. 1995). Species interactions were likely to be important, but their evaluation was
outside the scope of this study.
17
CHAPTER 2
EXPERIMENTAL DESIGN AND METHODOLOGY
INTRODUCTION
An overarching hypothesis of the VCR LTER project is that relative elevations of
land, sea, and freshwater table are expected to drive ecological processes. Furthermore
species composition on pimples appeared to change relative to land elevation on a sub-
meter scale. Because of those two factors, accounting for elevation in the sampling
scheme was important. Rather than sampling species along an elevation gradient, it was
more appropriate to limit sampling units to roughly the same elevation on the pimple.
That fact combined with the density of the shrub thickets on the pimples and the apparent
narrowness of vegetation zones provided cogent reasons for choosing a linear sampling
method instead of a quadrat or other two-dimensional technique. I tested two linear
sampling techniques: line-intercept and point-intercept (described below), for their
efficacy and statistical power in measuring species relative cover or density along the
same elevation (Godfrey et al. 1979, Bonham 1989, Dale 1999).
PILOT STUDY
Pilot sampling of species and tests of the two sampling technique occurred on
three Hog Island pimples in June–July 2002. Each sampling unit was a 5 m linear transect
marked with measuring tape and oriented parallel to the elevation contour of the pimple.
Transect placement was stratified along the elevation gradient; there were three transects
in each of four surveyed quadrants, which faced the cardinal directions, i.e., 12 transects
per pimple. Relative cover for each species was recorded for each transect using both the
point-intercept method, in which species contacting a rod or plum line at 50 random
points along a transect are recorded as well as the line-intercept method, in which linear
18
distances of plant coverage are recorded. For the first method, abundances would be
recorded as number of encounters, and for the second abundance would be recorded as
cm of coverage by each species. There were two sampling strata recorded for each
method: plant species less than one m tall belonged to the understory and woody species
above one m were in the shrub layer or overstory. I evaluated each method for ease and
expedience, compared results from each method, and used the point-intercept data to
estimate the number of point intercepts needed to confidently sample within 10% of the
true mean (Dale 1999).
In the field, the point-intercept method was more efficient than line-intercept
measurements because it reduced the total number of observations made per transect.
Paired t-tests determined if the percent cover values for species in transects differed
between the two sampling methods. Species cover recorded with the line-intercept
method did not differ from species cover recorded at 50 points along a transect (p = 0.96).
Furthermore, both random and stratified sub-samples of the point-intercept data did not
significantly differ from the original line-intercept data for sub samples containing at
least 10 of the original points (p = 0.74–0.98), suggesting that fewer points could be used
without sacrificing accuracy. Nevertheless, computing sample size estimation for the
point-intercept data revealed that > 100 points/transect would be required to sample
consistently within a 10 % confidence interval around the true mean (Bonham 1989). In
other words, statistical power would be lost with the more efficient sampling method.
Conversely, the line-intercept method provides the actual species abundances rather than
a sub-sampled mean and is not subject to concerns about statistical power.
19
During field trials in summer 2002 and 2003, a twelve-transect sampling array
proved prohibitively time-consuming for floristic surveys; i.e., it would not be possible to
finish an entire survey of all sites before growing conditions changed. Instead of twelve
transects, I decided to use a three-transect array.
FIG. 3. Sampling method.
FINAL SAMPLING CONFIGURATION
To establish a sampling array, I randomly located a survey point outside of a
pimple. Using an electronic transit (Pentax Total Station laser theodolite), I surveyed
three more transect center points in a straight line connecting the center of the pimple
with the survey point (FIG. 3). I stratified placement of center points by elevation. The
Peripheral slope with Morella
Flat summit with xeric forbs & grasses
5 m sampling transect centered on survey flag
median line of transects oriented on a random azimuth
Main survey point in marsh
Marsh zone around pimple
N
20
first transect was in marsh at the periphery of shrubs; the second transect was on the slope
of the pimple sides under shrubs; and the central transect was placed outside the shrub
zone in the pimple summit. It was not feasible or practical to survey elevation waypoints
on Hog Island in order to calculate true elevation above sea level; therefore the relative
elevation of each transect above marsh ground level, i.e., the surface of the muck layer
underneath the detritus or water, was recorded. The relative distances from the origin
point and the relative elevations allowed simple trigonometric calculation of slope at each
transect. Using GPS coordinates of the origin survey point and the transect center in the
pimple interior allowed a fairly accurate calculation of the azimuth (aspect) of the
transect array. Array installation on Parramore Island pimples differed from the above
methods in that a tripod-mounted laser level and stadia were used instead of the
automatic transit.
Permanent transects were established during summer 2003 for Hog Island pimples
and summer 2005 for Parramore Island pimples. Pimples on Hog Island lie along a nearly
North – South line in the oldest interior swale marsh between ridges that date from 1871
and 1955; most lie closer to the 1955 ridge. Seventeen pimples were chosen randomly
from a ca. 700 m section of that line centered on the vehicle trail (Fig. 4). Pimples on
Parramore Island are in the southernmost half of the island, an area fronted on the lagoon
side of the island by a ~ ½ km wide strip of soft-bottom salt marsh and tidal creeks that
are too shallow for motorized boats at low tides. This made reaching pimples on
Parramore impractical and potentially dangerous for most of the study. Four pimples
studied on Parramore Island were chosen based on accessibility from the Swash, a
channel through the marsh, and were roughly in a southwest–northeast line connecting
21
442641 E, 4151840 N and 443060 E, 4152222 N (UTM zone 18, WGS84 datum; latitude
– longitude equivalent is 37.512° N, 75.649° W to 37.515° N, 75.654° W).
FIG. 4. Map and aerial photograph of pimples and main dunes sampled on Hog Island. The area represented by the graphical map (A) is denoted by the rectangle on the aerial photograph (B). Open circles and squares in B respectively represent the location of the long-term swale and dune water table-monitoring wells. Coordinates are for UTM zone 18 using the WGS84 datum. The northwest and southeast corners of the map (A) are 75.670°W, 37.454°N and 75.667°W, 37.446°N, respectively.
11995555
11887711
22
During the same period that the pimple transects were being established on Hog
Island, I established ten more transects near two water-recording wells, S4 and S3, which
are located in swales associated with the oldest and second-oldest dune ridges on the
island, respectively (Conn and Day 1993, 1997, Dilustro and Day 1997, Day et al. 2001).
Not only are these wells of known elevation, they also provided water level data for
transects. Five more transects were set up near wells D3 and D4, which were located on
dunes near the swale wells. In addition to data from those transects, species data were
included from control plots used for a long-term nutrient addition study located in the
dunes around the wells (Stevenson and Day 1996, Conn and Day 1997, Day et al. 2001).
SPECIES COLLECTION
At the start of the study, I began collecting voucher specimens for species
encountered on and around the pimples of Hog Island and eventually all new species,
including ferns and mosses, and a few unique species from Parramore Island. The
specimens were identified and deposited at the Old Dominion University Herbarium.
23
CHAPTER 3
PLANT ASSEMBLAGE STRUCTURE AND DIVERSITY PATTERNS
INTRODUCTION
The basic goal of community ecology is to describe the interrelationships, social
structures, and environmental interactions of the species and habitats within a specified
region (Cowles 1899, Clements 1936, Whittaker 1956, 1960, Magurran 1988, Bonham
1989, Fauth et al. 1996, Morin 1999). Some of the earliest studies of biodiversity were in
dune plant assemblages (Cowles 1899, Clements 1936, Magurran 1988). Investigations
of dune and barrier island ecology are still relevant today because they advance
knowledge about general ecological theories and concepts like succession and island
biogeography. Understanding barrier islands also has practical implications for dune
restoration, wetland mitigation, and wildlife management (Hosier and Eaton 1980,
Ehrenfeld 1990, Johnson and Barbour 1990, Hayden et al. 1995, Erwin 1996).
Because dunes are frequently disturbed, the species composition of plant
assemblages on them can change significantly in short time spans and ‘climax’
communities are either unattainable or unstable (Ehrenfeld 1990, Young et al. 1994).
Studying the rapid species turnovers and system state changes may help ecologists
understand less dynamic or more slowly-changing ecosystems (Belsky and Amundson
1986, Ehrenfeld 1990, de Castro 1995, Anthonsen et al. 1996, Day et al. 2001). This may
especially be true for learning about the impact of edaphic and geomorphological factors
on the sequence of succession such as alterations in dune stability, height, disturbance
frequency, and soil amendments (both biotic and abiotic) (Fahrig et al. 1993, Young et al.
1994, Young et al. 1995a, Young et al. 1995b).
24
Island biogeography and the species – area relationship are basic concepts in
ecology (Gleason 1925, MacArthur et al. 1966, Connor and McCoy 1979). Nevertheless,
theories concerning relationships of isolation and habitat area to species richness and
diversity are still being developed and refined (Crist and Veech 2006). Barrier islands and
dune communities are well-suited to investigations on those topics because they are both
ecosystems of varying size and isolation.
Because most of the world’s population lives in coastal areas, understanding
coastal and dune ecosystems often has demonstrable, practical value in terms of
protecting coastal resources and real estate (Brinson 1996). Describing patterns of plant
diversity on dunes helps to advance knowledge about ‘healthy’ dune ecosystems and
remains important to the study of ecology (Clayton 1972, Christian et al. 1998).
Virginia’s barrier islands are excellent research subjects in this regard because they
represent some of the most pristine and most extensive coastal habitat on the US Atlantic
seaboard.
In this chapter, I describe the assemblage structure of pimples on Hog and
Parramore Islands using traditional density descriptors, three different ordination
methods, an assortment of diversity indices, and parametric tests. My intentions were to
provide a comprehensive study of the plant species on pimples and compare the
usefulness of some ‘traditional’ and newer techniques, with special emphasis on the
application of ordination and diversity indices.
General descriptors of assemblage structure
Basic mathematical methods used in describing patterns in communities include
species frequency, dominance, and importance; species–area curves; species
25
accumulation curves and rarefaction curves; and diversity indices (Curtis and McIntosh
1951, Dawson 1951, Whittaker 1960, McIntosh 1962, Bannister 1966, Will-Wolf 1980,
Magurran 1988, Palmer 2007). More elaborate methods that are becoming increasingly
popular since the advent of inexpensive computing power are ordination methods like
non-metric multidimensional scaling (Palmer 2007).
Besides species richness, which is the number of species, other basic descriptors
of assemblages are frequency, density, dominance, relative frequency, relative density,
relative dominance, and importance values (Curtis and McIntosh 1951, McIntosh 1962,
Bannister 1966, Will-Wolf 1980). These measures are good tools for a preliminary
overview of plant census data because they give an indication of the relative abundance
of a particular species. Nevertheless, relationships between species must be inferred, and
it is difficult to use importance values to reach an overarching assessment of assemblage
dynamics.
Rarefaction curves
Rarefaction curves are estimates of the cumulative number of species found as a
function of either the number of individuals sampled or the number of samples taken
(Olszewski 2004, Crist and Veech 2006). They allow interpolated estimates of total
species richness in a sample area and comparison of richness between two assemblages
that have not been sampled equally (Chao 1987, 1989, Colwell and Coddington 1994,
Lee and Chao 1994, Peterson and Slade 1998, Colwell et al. 2004, Olszewski 2004, Crist
and Veech 2006). This feature was important to me, since I was comparing pimple
transects with long-term transects that had been censused with different methods (Day et
al. 2001). Rarefaction curves also can be used to estimate sampling adequacy, species
26
diversity indices, and diversity partitions, when diversity is measured as species richness
(Chao 1987, Lee and Chao 1994, Chazdon et al. 1998, Crist et al. 2003). See the
discussion of diversity partitioning below.
Ordination
Ordination techniques are statistical methods usually employing sample averages
and matrix algebra, and most can be thought of as special adaptations of the general
linear model (Dawson 1951, Bray and Curtis 1957, Swan 1970, Gauch and Whittaker
1972b, Gauch 1973, Hill 1973b, Gauch et al. 1977, Gauch 1982b, Kent and Ballard 1988,
Crist et al. 2003, Palmer 2007). Their purpose is to reduce the number of dimensions in
multidimensional relationships by determining correlations among a set of research
objects and putting them in order along explanatory axes.(Bray and Curtis 1957,
Bannister 1966, Gauch and Whittaker 1972b, Gauch 1973, Gauch et al. 1977, Gauch
1982b, Kent and Ballard 1988, Clarke 1993, Palmer 2007). When plotted against two or
three explanatory axes, positions of research objects in the resulting ‘ordination’ space
reveals their relative similarity or dissimilarity (Palmer 2007). In theory, these patterns
are the result of species’ responses to environmental variation and interactions.
I had a choice of several ordination techniques that have risen and fallen in
popularity among different groups of ecologists over the years (Whittaker 1956, Bray and
Curtis 1957, Swan 1970, Walker and Wehrhahn 1971, Gauch and Whittaker 1972b,
Gauch 1973, Gauch and Chase 1974, Gauch et al. 1977, Peet and Loucks 1977, Orlóci
1978, Gauch and Stone 1979, Gauch 1982b, Pielou 1984, ter Braak and Barendregt 1985,
ter Braak 1986, Peet et al. 1988, Trejo-Torres and Ackerman 2002).While most
researchers agree that they should carefully choose the technique most appropriate to
27
their study, few agree on the circumstances under which a technique is appropriate
(Pielou 1984, Kent and Ballard 1988). I followed the example of other researchers and
evaluate the usefulness of a selection of techniques, rather than one, in describing pimple
assemblage structure (Wentworth 1981, Westman 1981, Chang and Gauch 1986). I used
two well-established techniques, Bray-Curtis ordination and principal components
analysis, and a more modern and computationally intensive method, non-metric
multidimensional scaling (Goodall 1954, Bray and Curtis 1957, McCune and Mefford
1999, Palmer 2007).
Diversity indices and diversity partitioning
Among all the ways to describe assemblage structure, biological diversity has
been a perennial theme of community ecology (Cowles 1899, Whittaker 1956, Bray and
Curtis 1957, Hill 1973a, Magurran 1988, Tilman 1997, Peltzer et al. 1998, Chiarucci et
al. 2001). Ecologists and environmentalists consider biological diversity to have intrinsic
value and to relate directly to ecosystem integrity (Magurran 1988, Colinvaux 1993,
Booth and Grime 2003). Researchers have been developing and refining mathematical
and field techniques for estimating diversity for a century or more (Cowles 1899, Gleason
1925, Bray and Curtis 1957, Gauch 1973, 1982a, Pielou 1984, Bonham 1989, Giannini
2003). Studying the diversity of assemblages on isolated geomorphological features like
pimple dunes presented a good opportunity to advance knowledge on this topic.
The three most common measures of species diversity are species richness (S),
Shannon’s diversity index (H), and Simpson’s diversity index (D). Both the Shannon and
Simpson’s indices take into account species richness as well as species evenness. I used
both indices because they are well-known and easy to calculate and interpret.
28
Partitioning diversity. I wanted to look for differences in assemblage structure
between habitat types (marsh, shrubs, and summits), between pimples and main dunes,
between islands, and among seasons. Fortunately, there are many theories and methods
for determining diversity at different spatial (and temporal) scales. A predominant view
first proposed by Whittaker (1966) is that diversity across a landscape can be broken into
at least three partitions: site-level diversity (α-diversity), the diversity that represents the
change in species between sites (β-diversity), and the total diversity of all sites in a
landscape (γ-diversity). Although some have found problems with this approach, e.g.,
defining the sizes of site and landscape, it has proven to be a useful concept with many
proposed methods of measurement (Whittaker 1960, Gauch and Whittaker 1972a, Alatalo
and Alatalo 1977, Shmida and Wilson 1985, Lande 1996, Whittaker et al. 2001, Veech et
al. 2002, Booth and Grime 2003, Summerville and Crist 2005, Crist and Veech 2006).
A problem with the partition concept of diversity is lack of comparability.
Whereas α- and γ-diversity are usually measured as richness or with a diversity index, β-
diversity is usually measured with any number of indices that are specific to it (Shmida
and Wilson 1985, Magurran 1988, Olszewski 2004). Although the utility of β-diversity
indices has been established (Shmida and Wilson 1985, Magurran 1988), they cannot be
directly compared to indices used to measure α- and γ-diversity. Without direct
comparability among all levels, the hierarchical partition concept is less useful. I decided
it would be instructive to calculate a few direct measures of β-diversity, but wanted to
find a method for measuring diversity that would be comparable at all scales: α, β, and γ.
Additive partitions. A recent advance in measuring partitions of diversity was
well-suited to the needs of my study. This development is the concept of Whittaker’s
29
partitions being additive, instead of multiplicative as he first proposed (Lande 1996,
Veech et al. 2002, Crist and Veech 2006). Within this framework, diversity partitions at
each level are additive, such that the mean diversity within samples (α) plus the mean
diversity among samples (β) equals the total system diversity (γ), regardless of the
statistic used to define diversity (Veech et al. 2002, Crist and Veech 2006). Furthermore,
there can be any number of hierarchical partitions of diversity based on scale, i.e.,
multiple α and β levels (Crist and Veech 2006). It has also been suggested that
hierarchical levels of diversity can include temporal levels as well as spatial ones
(Summerville and Crist 2002, Crist et al. 2003, Summerville and Crist 2005). I used
methods based on these theories to measure diversity at the transect, pimple, and season
level. There have not been many comprehensive studies of additive diversity partitions at
different spatial and temporal scales and at present none have been conducted in barrier
island or dune systems.
Rarefaction curves. Rarefaction curves (see above) also have use for determining
diversity partitions when diversity is measured as species richness. The last point on a
curve is Sobs or γ-level richness (Summerville and Crist 2002, Veech et al. 2002, Crist et
al. 2003, Summerville and Crist 2005, Crist and Veech 2006, Summerville et al. 2006).
Using rarefaction curves for estimating diversity partitions is informative because it
provides a graphical representation of the partitions.
Study goals
The principle goals of the current study were: 1) define assemblage groups on
different pimples, 2) estimate the total number of species on the pimples, 3) look for
evidence of succession in temporal variation, and 4) determine significance in spatial
30
patterns of diversity at different scales. My goal for examining spatial patterns was to
find A) a relationship between distance and similarity in pairwise comparison of pimples,
B) a discernible spatial relationship among species within pimples, and C) significant
differences in diversity at the transect, pimple, and island level, i.e., α-, β-, and γ-diversity
(Whittaker 1960, Gauch and Whittaker 1972a, Peterson and Slade 1998, Gotelli 2001,
Crist and Veech 2006). Where possible and practical, I evaluated more than one approach
to each of these questions for the sake of comparison.
MATERIALS AND METHODS
Field methods are described in Chapter 2. I used pivot tables in Microsoft Excel
to calculate the sum of species measurements within each transect. The spatial
distribution of species within transects is therefore not represented in my analyses.
General descriptors
Species richness, diversity indices, and parametric tests. In addition to recording
species richness, I calculated Shannon and Simpson’s diversity indices at the transect-,
pimple-, and whole-island level. Both indices account for species richness and species
evenness by using the proportions (pi) of total abundance represented by each species in
their calculation (Table 1). Conventionally, pi is calculated as the number of individuals
of species i divided by the total number of individuals for all species in the sample. The
range of values for the Shannon index varies relative to S (Hill 1973a, Magurran 1988,
Crist et al. 2003, Crist and Veech 2006). The term ‘Simpson’s index’ has been applied to
three different calculations: Simpson’s index, Simpson’s index of diversity, and
Simpson’s reciprocal index. Both of the first two measures produce values in the range of
0–1, but in the first index diversity is high as values approach zero; the opposite is true
31
for the second index. Values for the reciprocal index vary 1–S. The second index has the
advantages of having the same range of outcomes regardless of S and being intuitively
interpreted, i.e., diversity increases as values increase. Simpson’s index of diversity is
therefore the index used in my analyses (Hill 1973a).
TABLE 1. Calculation of Shannon and Simpson’s diversity indices (Hill 1973a, Crist and Veech 2006). Variables:
H = Shannon diversity index Dx = different variations on Simpson’s index S = species richness in sample pi = proportion of the abundance of the ith species relative to the total abundance
of all species; e.g., (# individuals of species i)/(total # individuals). Formulas:
Shannon index H p pi ii
S
= −=
∑ ln1
Simpson’s index, proper D pii
S
12
1
==∑
Simpson’s index of diversity D pii
S
22
1
1= −=∑
Simpson’s reciprocal index Dpi
i
S32
1
1=
=∑
I used ANOVA to test hypotheses that species richness and species diversity
varied at both transect and pimple level based on habitat type and year. For each test of
richness, Shannon diversity, and Simpson’s diversity, I used a main effects model with
year, habitat, pimple, and transect as terms. Full-factorial models did not allow enough
degrees of freedom for a test to be performed. I performed multiple comparisons with
Tukey’s test. I considered P ≤ 0.05 significant for both ANOVAs and multiple
comparisons. When the outcome would be biased by an incomplete dataset, I excluded
32
2005 data from calculation of annual means and most ANOVAs. I also calculated means
for species abundance data from the long term study on the main dunes as a point of
comparison.
Importance values. Individual counts allow more options with diversity analyses
and other descriptors than continuous measures of species abundance, such as % cover.
Nevertheless, I did not attempt to count individual ramets because the majority of species
were clonal and any individual count would have been artificial and prone to bias. An
example of problems arising from this situation involves the calculation of importance
values (IVs) for species (Table 2).
TABLE 2. Species abundance measures used in calculating species importance values (Curtis and McIntosh 1951, McIntosh 1962, Bannister 1966, Will-Wolf 1980).
Measurement calculation
number of plots in which a species is found Species frequency total number of plots surveyed
number of individuals found for a species in all plots Species density total area (or other measure of sampling effort) surveyed)
total coverage recorded for a species Species dominance total area or distance surveyed (i.e. total potential coverage)
frequency for a given species Relative frequency sum of all species frequencies
density for a given species Relative density sum of all species densities
dominance for a given species Relative dominance sum of all species dominance values
Importance Value
relative frequency + relative density + relative dominance
33
I calculated a yearly average of dominance, relative dominance, frequency, and
relative frequency for each species, but, without counts of individuals, I could not
calculate density or relative density. As an alternative, I included IVmod (my designation)
for species, which is the sum of relative dominance and relative frequency.
Rarefaction curves
Calculation of the expected outcome curves can be accomplished both by simulation and
with analytical formulas, the most common being Mao’s tau, τ (Table 3) (Chao 1987,
1989, Lee and Chao 1994, Chazdon et al. 1998, Olszewski 2004). I used the EstimateS
8.0 program to produce rarefaction curves, Mao’s tau values, and the two rarefaction-
curve species richness indicators used in my analyses: Chao 2 and incidence-based
coverage estimator (ICE) (Colwell et al. 2004, Crist and Veech 2006). These estimators
use the incidence of rare species and the total observed species richness (Sobs) to predict
the actual S (Chao 1987, Lee and Chao 1994, Chazdon et al. 1998, Crist et al. 2003). By
determining if the curve has reached an asymptote or by comparing Sobs to a richness
estimator, one can assess how successful field sampling was in capturing all species
present. Confidence intervals for the range of likely total richness values can also be
calculated, and Mao’s tau can also be used to predict confidence intervals along the entire
curve as per Colwell et al. (2004). I also used rarefaction curves to estimate diversity
partitions (see below).
34
TABLE 3. Formulas for species richness estimators that approximate rarefaction curves. The different sources of these formulas in part used the same letters for different variables. Throughout the dissertation, I changed variable designations from the originals to avoid confusion and maintain continuity (Chao 1989, Chazdon et al. 1998, Crist et al. 2003, Crist and Veech 2006). Variables:
Sobs = observed species richness for entire study Sk = the number of species found in exactly k samples (transects); e.g. S1 is the
number of species in one sample, S2 is the number of species in two samples, etc.
SChao2 = Chao2 true richness estimator Sice = incidence-based coverage estimator of true species richness Sfreq = number of species found in > 10 samples, transects, or quadrats Sinfr = number of species found in ≤ 10 samples, transects, or quadrats B = the total number of samples (transects) b = the number of samples in a subset; b = 1,2,3…B Qk = number of species that occur in k samples, transects, or quadrats; e.g., Q1
= the number of species found in only one sample γice = estimated coefficient of variation for Qk of infrequent species (Chazdon et
al. 1998) Cice = sample incidence coverage estimator (Chazdon et al. 1998) τ(b) = Mao’s tau unbiased estimator of species richness in b samples ξ = a combinatorial coefficient used in the calculation of τ(b) σ = a variance estimator for calculating the confidence interval of τ(b) S~ = unknown species richness used in calculating the confidence interval of
τ(b) Formulas:
Total observed richness: ∑=
=B
kkobs SS
1
Chao2: 2
21
2 2QQ
SS obsChao +=
ICE: ice
icerfreqice C
QSSS
21inf γ+
+=
Mao tau rarefaction estimator: k
B
kkbobsk
B
kkb SSSb ∑∑
==
−=−=11
)1()( ξξτ
where ⎪⎩
⎪⎨⎧
−−−−
=0
!)!()!()!(
BkbBkBbB
kbξ )(
)(
Bbk
Bbk
>+
≤+
note that when b = B, ξ = 0; therefore τ(B) = Sobs
35
(Table 3, continued)
Mao tau confidence intervals, i.e., τ(b) ± 1.96σ(b):
SbSb k
B
kkb
~/)()1()( 2
1
22 τξσ −−= ∑=
where 2
21
2)1(~
BSSB
SS obs−
+=
Ordinations
To describe communities with respect to species and habitat associations I used
multivariate ordination tests: Bray-Curtis ordination, non-metric multi-dimensional
scaling, and principal components analysis. I performed all analyses with PC-ORD v. 4;
the NMS analysis was preceded by an ‘autopilot’ analysis to determine the best number
of axes to ordinate the result. Pearson’s r correlation coefficient was used to judge the
variables most important to determining axes.
Resampling and diversity partitions
To examine the spatial partitioning of diversity and to determine if diversity
changed between years in each pimple, I used resampling simulations (PARTITION
program). I compared diversity partitioning between pimples and the main dunes and
compared those results with direct calculations of different indices of α-level (transect)
and β-level (turnover between transects) species diversity (Magurran 1988, Crist and
Veech 2006).
I used sample-based randomizations to calculate diversity partitions rather than
individual-based randomization. Because sample-based randomization “preserves the
36
patterns of intraspecific aggregation in the observed data, it is most useful in testing
explanations of species diversity that are based on nonrandom species assemblages”
(Crist et al. 2003). To perform a sample-based randomization analysis of diversity
partitions, values from observed samples at level h are randomly allocated to the larger
sample units at level h + 1 that all belong to the same even larger sample unit at level
h + 2. This process is repeated 1,000 – 10,000 times, typically, to create a bank of null-
value data sets, i.e., diversity values that would occur if samples were randomly
distributed within each sampling level. The proportion of null values that exceeds a
particular observed value then becomes the probability (P-value) that a greater value than
the observed would occur by chance. Thus, very high probabilities (P ≥ 0.95) indicate
that observed values were considerably less than random values and should be considered
statistically significant in addition to those probabilities ≤ 0.05.
I used three different indices of β-diversity (Table 4). Whittaker’s measure of β-
diversity relies on the ratio of total species richness to average sample richness
(Whittaker 1960). Cody’s measure uses species turnover between sites (Magurran 1988).
Wilson and Shmida’s measure combines Whittaker’s with Cody’s (Wilson and Shmida
1984, Shmida and Wilson 1985, Magurran 1988).
Direct measurements of diversity partitions were made using the concept that
partitions are additive. Consider, for example, a hypothetical case with four partitions of
diversity: the average α-level diversity in each transect, the average β1-level between
transects, the average β2-level between plots, and the overall diversity (γ). In that case,
α1+β1=β2 and α1+β1+β2=γ (Table 5); this is true if diversity is measured as species
richness (intuitive), or with either Shannon’s or Simpson’s diversity index (perhaps not as
37
intuitive; Crist and Veech 2006). Theoretical ecologists have devised methods to
capitalize on the additive property of diversity partitions and calculate α- and γ-diversity
so that β-diversity levels can be determined as the difference between them (Table 5)
(Whittaker 1960, Gauch and Whittaker 1972a, Magurran 1988, Colwell and Coddington
1994, Lande 1996, Chazdon et al. 1998, Peterson and Slade 1998, Gotelli 2001, Longino
et al. 2002, Veech et al. 2002, Colwell et al. 2004, Olszewski 2004, Crist and Veech
2006). Repeated measurements of diversity can be regarded as making up the highest
intermediate level of diversity (βm).
TABLE 4. Three measures of β-diversity (Whittaker 1960, Wilson and Shmida 1984, Shmida and Wilson 1985, Magurran 1988).
Variables: βW = Whittaker’s measure βC = Cody’s measure βT = Wilson and Shmida’s measure Sobs = total species richness found in the system α = mean sample-level diversity, where diversity is measured as species-richness G = number of species gained along a series of samples L = number of species lost along a series of samples Formulas: Whittaker’s measure: βW = −( / )Sobs α 1 Cody’s measure: βC = (G + L) / 2 Wilson and Shmida’s measure: βT = (G + L) / 2α
As an alternate to the resampling method of partition estimation I also used
rarefaction curves. The first point on the curve is mean richness at the first level: α1
richness. Following the formula ∑=
+=m
hh
11 βαγ (Table 5), the difference between the first
and last points on the curve is the total β-richness for the system. Furthermore, in a
second rarefaction curve that estimates accumulation of species by pimple instead of
38
transect, the first point on that curve would be the mean α-level richness per pimple. The
difference between the first point on the pimple-level curve and the first point on the
transect-level curve equals β-level diversity among transects (Crist and Veech 2006).
That is, α2 – α1 = β1; this relationship can be used with curves at each successive level to
determine α- and β-level richness (Gotelli 2001, Longino et al. 2002, Colwell et al. 2004,
Olszewski 2004, Crist and Veech 2006).
RESULTS
General descriptors
Pimple-level patterns. The mean number of species found in each pimple across
all years of the study was 9.7 ± 4, and annual mean species richness in pimples was
significantly higher in 2006 than in either 2003 or 2004 (ANOVA; P < 0.01; Fig. 5).
There were significant differences in pimple-level diversity between pimples (ANOVA;
richness: P < 0.001; Shannon: P < 0.001; Simpson’s P < 0.02; Fig. 6). Richness varied
between pimples by a factor of three, and Shannon and Simpson’s diversity indices also
varied by an appreciable margin (FIG. 6). Variance in diversity does not appear to be
determined by pimple size or location (FIG. 6), and I verified that observation with
regression analysis (Fig. 7).
I performed regression analyses on the response of species richness to 1) different
estimates of pimple size (e.g. maximum width east to west or north to south or maximum
radius; Fig. 7A); as well as 2) different measures of pimple location (i.e. x or y coordinate
on a map grid; Fig. 7B). Regression results did not indicate any significant relationships;
R2 values were < 0.10 and coefficients were < 0.001. Similarly, there was no relationship
between either Shannon or Simpson’s diversity indices and pimple size and position.
39
TABLE 5. Terms and formulas used in additive diversity partitioning (Crist et al. 2003, Olszewski 2004, Crist and Veech 2006). .
Variables: γ = total species richness or diversity in study area; γ < αm αh = average richness or diversity within sample units at level h m = the highest level of study h = 1, 2, 3,…m are all the levels of study βh = average richness between sample units at level h rh = the number of sample units at level h j = the sample number at a particular level, i.e., j = 1,2,3…rh bhj = the jth sample at level h Dhj = the diversity metric recorded for the jth sample at level h qhj = the proportion of the number of individuals in sample j to the total number
of individuals in all samples at level h Formulas:
total (landscape) diversity ∑=
+=m
hh
11 βαγ
average diversity at level h; unequal sample weights:
∑=
=hr
jhjhjh qD
1α
average diversity at level h; equal sample weights (used in estimating partitions
from rarefaction curves):
∑=
=hr
jhj
hh D
r 1
1α
general equations for diversity among samples or higher levels (β-diversity) 1) βm = γ – αm
2) βh = αh+1 – αh ; for h<m
average diversity among samples at level h when D is measured as species richness; unequal sample weights
)(1
hj
r
jhjh Dq
h
−= ∑=
γβ
average diversity among samples at level h when D is measured as species richness; equal sample weights
)(11
hj
r
jhh D
r
h
−= ∑=
γβ
40
2003 2004 2005 2006 TOTAL0
4
8
12m
ean
spec
ies
richn
ess
in p
impl
es
2003 2004 2005 2006 TOTAL
1
2
3
4
5
mea
n sp
ecie
s ric
hnes
s in
tran
sect
s
FIG. 5. Per-pimple and per-transect mean species richness by year with grand total mean. In this and subsequent figures, error bars represent one standard error of the mean and lowercase letters represent significantly homogeneous groupings (at p < 0.05) based on Tukey’s multiple comparisons test. The incomplete dataset from 2005 was not included in multiple comparisons.
FIG. 6. Maps showing pimple sizes, richness, Shannon diversity, and Simpson’s diversity. In the area map, dot size reflects area. In the diversity maps, relative positions of pimples are preserved, but the diameters of symbols indicate the relative values of each diversity metric. Numbers indicate maxima and minima for estimation of scale. Color patterns represent overlapping homogeneous subsets as determined with Tukey’s test.
a a b
5.5
16
0.80
0.79
0.43
1.95
a a b
1955 thicket
pond
pond
1871 thicket
trail
41
FIG. 7. Relationships of pimple size and location to richness.
Transect-level patterns. Mean richness per transect was 4.3 ± 2.5 (Fig. 5). There
was a significant effect of habitat type on species richness, Shannon diversity, and
Simpson’s diversity index; marsh and summit transects had higher mean values than
shrub transects (ANOVA; P < 0.001 for all three measures; Tukey’s significant at
P < 0.05; Fig. 8, Fig. 9, & Fig. 10). Year of measurement had a significant effect on
variation for each of the three metrics (ANOVA; P < 0.001 for all three measures;
Tukey’s significant at P < 0.05), but only the incomplete 2005 dataset was significantly
different from other years. Analysis with 2005 data excluded revealed that there was no
significant difference in species richness between 2003, 2004, and 2006, but there was a
significant year effect with Shannon and Simpson’s diversity with 2004 transects
exhibiting higher diversity than the other two years (ANOVA; P < 0.05 for both metrics;
Tukey’s significant at P < 0.05).
0
4
8
12
spec
ies
richn
ess
A A A A A
A A A A A A A A A A A AA A A A A A A A A A A A A A A A A A A A AAAAA AAA A A A A A A A A A AAA AA AAA A A A
A A A A A A AAAAA A
A A A A A A A AAAAAAAA AA
A A A AAAAAA A A
A AA
A A
A A A
richness = 13.20 + -0.00 * coordinateR2 = 0.02
B north-south coordinate (m)
500 1000 1500
pimple area (m2)
0
4
8
12
spec
ies
richn
ess
AA A A A A A A A A
A A A AAAAA A A A A A A A A A A A A A A A A A A A A A A A AAA A A A A A A A A A A AAA A A A A A A A A A A A A A A A A A A A A A A AA A A A A A A A A A A A A A A A A A A A
A A A A A A A A A AAA A A A A A A A A A A A AAAAAA AA A A A A A A A A A A A A A A A A A AA A A A A A A A A
AA A
A
richness = 2.59 + 0.00 * areaR2 = 0.12
A 4600 5400
42
FIG. 8. Yearly mean transect-level species richness in each habitat type.
FIG. 9. Yearly mean transect-level Shannon diversity in each habitat type.
FIG. 10. Yearly mean transect-level Simpson’s diversity index in each habitat type .
2003 2004 2005 2006 TOTAL0.0
0.2
0.4
0.6
Sim
pson
’s d
iver
sity
marshshrubsummit
a a
b
2003 2004 2005 2006 TOTAL
0
2
4
6 a
a
b
marshshrubsummit
spec
ies
richn
ess
marshshrubsummit
2003 2004 2005 2006 TOTAL0.0
0.5
1.0
1.5
b
a a
Shan
non
dive
rsity
43
Comparison to main dunes. Mean richness on the main dunes was 19.5 ± 7.5 per
dune, 6.8 ± 2.5 per 25 m2 plot, and 3.8 ± 1.6 per .25 m2 subplot. The richness of pimple
summit transects at 5.1 ± 3.1 falls between the plot and subplot richness of the main
dunes (Fig. 11: B & C). Dune ridge reference plots did not have significantly different
Shannon diversity than summit plots on pimples (1.05 vs. 0.98; p = 0.2), but did have
significantly lower species richness (3.8 vs. 5; p < 0.0001).
Importance values. There were more infrequently-occurring species in pimple
plots than commonly-occurring ones (Fig. 12; Table 6). The most commonly-occurring
species was wax myrtle, Morella cerifera, which was also the most dominant (i.e., it
represented the most coverage.) The ten most dominant and frequently-occurring species
were all typical of shrub thicket or marsh. Spartina patens, a C4 grass, is commonly and
generally found in both hydric and xeric habitats on Hog Island, but is infrequent on
pimple summits. A few species were not found often but were very abundant when they
occurred, e.g. duckweed, Lemna minor.
FIG. 11. Rarefaction curves of species encountered based on number of transects sampled. The curves are the product of an analytical procedure (Colwell et al. 2004) and are therefore not empirically-derived species–area curves. (A) Entire pimple richness. (B) Dune reference plots. (C) Pimple summits.
200150100500
70
60
50
40
30
20
10
0
# sp
ecie
s
95% C I Upper Bound95% C I Low er BoundObserved Species
# transects4002000
95% C I U pper Bound95% C I Low er BoundObserved S pecies
# transects6050403020100
95% CI upper bound95% CI lower boundObserved Species
# transects
A B C
44
TABLE 6. Relative dominance and relative frequency for pimple species. “ IVmod ” is a modified importance value: relative frequency plus relative dominance. Species IVmod rank Rel. Freq. rank Rel. Dom. 1 Morella cerifera 75.10% 1 20.03% 1 55.07% 2 Polygonum hydropiperoides 16.00% 3 8.12% 2 7.87% 3 Parthenocissus quinquefolia 12.51% 2 8.96% 6 3.55% 4 Spartina patens 11.66% 4 6.16% 3 5.50% 5 Distichlis spicata 9.67% 6 4.62% 4 5.05% 6 Scirpus pungens 6.62% 9 2.94% 5 3.68% 7 Rubus argutus 6.48% 8 3.92% 7 2.55% 8 Carex albolutescens 6.27% 5 5.04% 10 1.23% 9 Mikania scandens 5.28% 7 4.06% 11 1.22% 10 Festuca rubra 5.02% 10 2.80% 8 2.22% 11 Rumex acetosella 4.05% 11 2.66% 9 1.39% 12 Juncus dichotomus 3.52% 12 2.38% 12 1.14% 13 Schizachyrium scoparium 2.86% 14 1.82% 14 1.04% 14 Baccharis halimifolia 2.81% 16 1.68% 13 1.13% 15 Cirsium horridulum 2.78% 13 2.10% 18 0.67% 16 Hydrocotyle verticellata 2.36% 19 1.54% 16 0.82% 17 Phragmites australis 1.98% 18 1.54% 20 0.44% 18 Cyperus strigosus 1.96% 22 1.26% 17 0.70% 19 Eupatorium hyssopifolium 1.85% 21 1.26% 19 0.59% 20 Prunus serotina 1.78% 15 1.68% 31 0.10% 21 Phyla lanceolata 1.74% 17 1.54% 27 0.20% 22 Panicum sp. 1.53% 20 1.40% 28 0.13% 23 Ammophila breviligulata 1.46% 24 1.12% 22 0.34% 24 Andropogon virginicus 1.38% 25 0.98% 21 0.40% 25 Persea palustris 1.22% 23 1.12% 32 0.10% 26 Lemna minor 1.19% 42 0.28% 15 0.91% 27 Juniperus virginianus 0.96% 28 0.70% 25 0.26% 28 Toxicodendron radicans 0.90% 27 0.84% 39 0.06% 29 Panicum lanuginosum 0.88% 26 0.84% 41 0.04% 30 Eupatorium capillifolium 0.70% 36 0.42% 23 0.28% 31 Teucrium canadense 0.63% 31 0.56% 35 0.07% 32 Typha latifolia 0.63% 30 0.56% 36 0.07% 33 Galium sp. 0.61% 29 0.56% 40 0.05% 34 Setaria geniculata 0.55% 41 0.28% 24 0.27% 35 Typha angustifolia 0.54% 35 0.42% 29 0.12% 36 Panicum amarum 0.50% 40 0.28% 26 0.22% 37 Pluchea odorata 0.50% 34 0.42% 33 0.08% 38 Panicum leucothrix 0.44% 33 0.42% 43 0.02% 39 Hypericum hypericoides 0.43% 32 0.42% 46 0.01% 40 Iva frutescens 0.38% 39 0.28% 30 0.10% 41 Fimbristylis caroliniana 0.29% 38 0.28% 45 0.01% 42 Linaria canadensis 0.29% 37 0.28% 47 0.01% 43 Euthamia capillifolium 0.22% 52 0.14% 34 0.08% 44 Ptilimnium capillaceum 0.21% 51 0.14% 37 0.07% 45 Panicum virgatum 0.21% 50 0.14% 38 0.07% 46 Samolus valerandi 0.16% 49 0.14% 42 0.02% 47 Hypochaeris radicata 0.16% 48 0.14% 44 0.02% 48 Boehmeria cylindrica 0.15% 46 0.14% 49 0.01% 49 Vitis sp. 0.15% 47 0.14% 48 0.01% 50 Panicum dichotomiflorum 0.14% 45 0.14% 50 0.01% 51 Kosteletskya virginica 0.14% 44 0.14% 51 0.01% 52 Juncus biflorus 0.14% 43 0.14% 52 0.01%
45
In main dune plots, relative dominance and frequencies of species were notably
different than in pimples, even when comparing the dune species only to transects from
pimple summits (Table 7). Although there were 38 species found in both the reference
and pimple summit plots, only 20 were shared between the two areas. The two most
important species in the reference plots were dune-stabilizing grasses, and wax myrtle
was only twentieth in importance rather than first, as in the pimple plots. In summit plots
on pimples, shrub and marsh species were common (e.g. Polygonum hydropiperoides,
Parthenocissus quinquefolia, and Rubus argutus), but infrequent or absent in dune plots.
Rarefaction curves
There were 52 species found in pimples during all four years of the study (Fig.
11A). The ICE and Chao 2 estimates of species richness were respectively 59.9 and 58.4
(Fig. 13). There were 38 ± 9.2 (95% CI) species found in the main dunes, compared with
35 ± 4 in the summit plots on pimples (Fig. 11 B&C).
Ordinations
Three different ordination techniques produced results that further established the
difference between pimple plant assemblages and species composition in reference plots
especially when comparing xeric transects, i.e., main dune and pimple summits. In both a
Bray-Curtis ordination (Fig. 14) and a non-metric multidimensional scaling procedure
(Fig. 15) describing relationships between transects, a few species were recurrently
highly correlated to explanatory axes (Table 8). Those species included wax myrtle,
(Morella cerifera); beach grasses (Ammophila breviligulata and Panicum amarum);
xerophytic forbs (Solanum carolinianum, Cirsium horridulum, and Rumex acetosella);
46
hydrophytes (Polygonum hydropiperoides and Carex albolutescens); and woody species
(Prunus serotina and Parthenocissus quinquefolia).
TABLE 7. Comparison of species importance value (modified), frequency, and dominance scores for species found in the main dunes of Hog Island and the summits of pimples. Species are listed in decreasing order of importance on the main dunes. Only importance value rankings are given; the modified importance value is the sum of relative frequency and relative dominance. Main Dunes Pimple Summits
Species IVmod rank
relative frequency
relative dominance
IVmod rank
relative frequency
relative dominance
Ammophila breviligulata 1 1 0.194 1 0.340 13 12 0.028 13 0.011 Panicum amarum 2 2 0.172 2 0.170 20 24 0.007 16 0.007 Spartina patens 3 3 0.109 3 0.093 4 7 0.045 3 0.062 Rumex acetosella 4 4 0.088 6 0.046 3 4 0.066 4 0.044 Schizachyrium scoparium 5 8 0.037 4 0.061 8 9 0.045 7 0.033 Cirsium horridulum 6 5 0.060 7 0.033 9 5 0.052 9 0.021 Rubus argutus 7 11 0.028 5 0.048 2 3 0.069 2 0.079 Panicum sphaerocarpon 8 9 0.037 8 0.031 - - - - - Eupatorium hyssopifolium 9 6 0.040 10 0.022 11 11 0.031 11 0.019 Prunus serotina 10 7 0.040 9 0.022 17 17 0.024 24 0.002 Solanum carolinianum 11 10 0.037 13 0.015 - - - - - Eupatorium capillifolium 12 14 0.016 11 0.021 19 19 0.010 14 0.009 Juncus dichotomus 13 13 0.020 12 0.016 7 6 0.049 6 0.035 Aristida tuberculosa 14 12 0.021 16 0.009 - - - - - Festuca rubra 15 15 0.013 15 0.011 6 8 0.045 5 0.041 Aralia spinosa 16 16 0.010 14 0.013 - - - - - Panicum lanuginosum 17 17 0.009 21 0.004 23 22 0.010 27 0.001 Mikania scandens 18 19 0.006 18 0.005 16 14 0.028 19 0.004 Morella cerifera 19 26 0.004 17 0.007 1 1 0.174 1 0.536 Baccharis halimifolia 20 18 0.006 20 0.005 30 30 0.003 21 0.003 Apocynum cannabinum 21 27 0.004 19 0.005 - - - - - Linaria canadensis 22 20 0.005 25 0.002 27 27 0.007 33 0.000 Centella erecta 23 21 0.005 24 0.002 - - - - - Cyperus sp. 24 22 0.005 26 0.002 - - - - - Gnaphalium chileensis 25 30 0.003 22 0.003 - - - - - Strophostyles helvola 26 23 0.005 27 0.002 - - - - - Linum medium 27 28 0.004 23 0.002 - - - - - Gnaphalium purpurea 28 25 0.004 28 0.002 - - - - - Fimbristylis caroliniana 29 24 0.004 30 0.001 35 35 0.003 32 0.000 Parthenocissus quinquefolia 30 29 0.004 29 0.001 5 2 0.087 10 0.020 Linum virginianum 31 31 0.003 32 0.001 - - - - - Monarda punctata 32 33 0.002 31 0.001 - - - - - Conzya (Erigeron) canadensis 33 32 0.002 34 0.001 - - - - - Hypericum hypericoides 34 35 0.002 33 0.001 29 29 0.007 38 0.000 Elymus virginianus 35 34 0.002 35 0.001 - - - - -
47
(Table 7, continued)
Species IVmod rank
relative frequency
relative dominance
IVmod rank
relative frequency
relative dominance
Euphorbium ammonoides 36 36 0.001 36 0.000 - - - - - Juncus canadensis 37 37 0.001 37 0.000 - - - - - Lepidium virginianum 38 38 0.001 38 0.000 - - - - - Polygonum hydropiperoides - - - - - 10 15 0.024 8 0.028 Carex albolutescens - - - - - 12 10 0.038 17 0.006 Andropogon virginicus - - - - - 14 16 0.024 12 0.013 Panicum sp. - - - - - 15 13 0.028 18 0.004 Juniperus virginianus - - - - - 18 18 0.017 15 0.008 Phragmites australis - - - - - 21 20 0.010 20 0.003 Toxicodendron radicans - - - - - 22 21 0.010 25 0.001 Persea palustris - - - - - 24 23 0.010 28 0.001 Scirpus pungens - - - - - 25 25 0.007 26 0.001 Panicum leucothrix - - - - - 26 26 0.007 30 0.000 Hydrocotyle verticellata - - - - - 28 28 0.007 34 0.000 Euthamia capillifolium - - - - - 31 31 0.003 22 0.002 Panicum virgatum - - - - - 32 32 0.003 23 0.002 Hypochaeris radicata - - - - - 33 33 0.003 29 0.001 Teucrium canadense - - - - - 34 34 0.003 31 0.000 Boehmeria cylindrica - - - - - 36 36 0.003 35 0.000 Vitis sp. - - - - - 37 37 0.003 36 0.000 Panicum dichotomiflorum - - - - - 38 38 0.003 37 0.000
Fig. 12. Histogram of pimple species frequency.
# encounters25 50 75 100 125
2
4
6
8
10
# Sp
ecie
s
48
200150100500# transects sampled
60
50
40
30
20
10
0
# sp
ecie
s
FIG. 13. Rarefaction curve showing partitions of diversity, observed species richness (Sobs) with confidence intervals, and two richness estimators: ICE and Chao 2.
Principal components analysis performed worse than the other two ordination
methods in explaining variation and resolving relationships, based on percent variation
explained by each axis (Fig. 16). Transects were crowded in the PCA ordination space,
apparently because of a few dune reference plots and the data from five Parramore Island
pimples. Removing these transects did not appreciably improve the data resolution. In all
ordinations, groupings of transects especially dune plots and pimple summits are distinct.
α pimples=9.3 β between transects=4.9
β between pimples=21.7
βbetween years = 22
γ = Sobs = 52
α within transects =4.4
α all pimples =31 (annual mean)
observed species
observed species:lower 95% CIobserved species:upper 95% CI Chao2 estimate ICE estimate
49
0.8
0.2 0.4 0.6 0.8
AXIS3
0.0 0.2 0.4 0.6
B
A
G
B
B
B
B
B
B
B
B
B
B
G AA
G
A
A
A
G
AB
A
A
B
G
A
A
A
AA
B
G
A
B
A
G
G
A
A
BA
A
AA
B
B
B
B
B
B B
B
B
B
B
B
B
B
BBB
B
B
B
B
B
B
B
AAA
A
A
A
GA
B
ABG
A
BGA B
G
A
B
G
A
BG
A
BG
A
B
A
B
G
A
BGA
B
GG
A
B
G
A
B
G
A
B
G AB
G
A
B
G
A
B
0.0
0.2
0.4
0.6
0.8
AXI
S2
0.80.0
0.0
0.3 0.50.8 1.0
G
BBB
B
B
BB
B
B
B
B
B
B
GGBGBG
A
BGBGBGBG
B
B
GG
B
GB
G
BGB
0.0
0.0
0.2
0.4
0.6
0.8
AXI
S2
0.0 0.3 0.50.8 1.0AXIS1 0.0 0.2 0.4 0.6 0.8
AXIS3
0.0 0.3 0.50.8 1.0AXIS1
0.3 0.50.8 1.0 0.0 0.2 0.4 0.6
B
G
BB
B
B
B
B
B
BB
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
A
A
G
A
B
AB
GA
BGA
B
G
A
B
B G
A
BG
A
B
G
A
B
A
B
0.0
0.2
0.4
0.6
0.8
B
BBB
B B
BBB
B
B B
B
B
B
BB
B
B
B
B
B
B
G
A
B
A
BGABG
A
BG
A
BG
A
BG
A
B
G
A
B
A
B
G
A
B
GA
B
GG
A
B
G
A
B
G
A
B
G
A
BG
A
B
0.0
0.2
0.4
0.6
0.8
A
G
B
2003 2004
2005 2006
Hog Isl. DunesHog Isl. PimplesParramore Isl.
marshshrubsummit
FIG. 14. Three axes of a Bray-Curtis ordination of transects based on species abundances. Axes 1-3 explain 37%, 7%, and 11% of variation in the data, respectively. Ordination was performed on data from all years combined; the plot was divided by year to increase clarity.
50
1.0
0.0 0.5 1.0
AXIS3
-0.5 0.0 0.5
B
A
B
B
BB
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
A
G
A
G
A
A
A
B
AAA
B
A
B
A
B
G
B
B
A
B
A
A
A
B
B
B
BBB
B
B
B
BB
B
BB
B
B
BB
B
B
BB
B
B
A
A
A
A
A
A
G
A
B
A
B
G
AB
G
A
B
G
A
B
G
A
BG
A
B
G
A
B
A
B
G
A B
GA
BG
G
A
B
G
A
B
G
A
BG
A
B
GA
B
G
A
B
-1.0
-0.5
0.0
0.5
1.0
AXI
S2
1.0 -1.0
-0.5
0.01.0
G
G
G
AGG
A
B
G
BG
A
B
GA
B
G
A
B
G
A
BGB
G
GG
A
G
A
G
GB
G
BGB
-1.0
-1.0
-0.5
0.0
0.5
1.0
AXI
S2
-1.00.0
1.0AXIS1 -0.5 0.0 0.5 1.0
AXIS3
-1.00.0
1.0AXIS1
0.01.0 -0.5 0.0 0.5
B
G
B
B
B
BB
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
A
A
G
A
B
A
B
G
ABGA
B
G
A
B B
G
A
B
G
A
B
G
A
B
A
B
-1.0
-0.5
0.0
0.5
1.0
B
B
BB
B
B
B
BB
B
B
BB
BB
BB
B
B
B
B
B
B
G
A
B
A
B
GA
B
GA
B
G
A
BG
A
BG
A
BG
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
A
B
G
A
B
G
A
BG
A
B
-1.0
-0.5
0.0
0.5
1.0
A
G
B
2003 2004
2005 2006
Hog Isl. DunesHog Isl. PimplesParramore Isl.
marshshrubsummit
FIG. 15. Non-metric multidimensional scaling of transects based on species abundances. An autopilot procedure determined that a three-axis ordination would provide the maximum amount of explanatory power. Axis 1 explains 25% of variation; Axis 2 21%; and Axis 3 36%. Ordination was performed on data from all years combined; the plot was divided by year to increase clarity.
51
0.0
-15.0-10.0-5.0 0.0
AXIS3
-20.0-15.0-10.0-5.0
B
AG
G
BB
BB
BBB
BB
B
B
BB
BB
B
B
BB
B
B
B
B
G
A
G
A
GA
A
AG R45a06
R47a06
R48a06
s41o06
G
A
B
A
B
GABGA
BG
A
BGABG
A
BG
A
BA
BG
A
B
G
A
B
G
G
A
B
G
A
B
G
A
B
G
A
B
GABG
AB
h08o06
B
BB
BB
B
B
BBBBBBBBBBB
BB
B
B
B
B
AAA
AA
A
R45a04
R48a04
G
A
B
ABGABG
A
B
GABG
A
BGA
B
G
A
BAB
G
A
B
GAB
GGA
B
GA
B
GA
B
GABGABGA
B
-10.0
-5.0
0.0
5.0
10.0
AXI
S2
0.0 -5.0
-20.0
0.05.0
10.0
-5.0
-10.0
-5.0
0.0
5.0
10.0
AXI
S2
-5.00.0
5.010.0AXIS1 -20.0-15.0-10.0-5.0 0.0
AXIS3
-5.00.0
5.010.0AXIS1
0.05.0
10.0-20.0-15.0-10.0-5.0
BG BBB
BBB
BB
BBBB
BBBB
B
BBB
B
BB
B
A
A R45a05 R48a05
s41r05
GABAB
G
ABGABGAB
B
GA
B
G
A
B
G
A
B
A
B
p02m05
p03m05
p03t05
p04o05-10.0
-5.0
0.0
5.0
10.0
BB
BBB
BB
BB
BBBB
BBBB
B
B
B
BB
B
R45a03R46a03R48a03
GABABGABGABGABGABGABG
A
BA
BG
A
B
GABGGA
B
GA
B
GA
BG
A
B
GAB
H10t03
-10.0
-5.0
0.0
5.0
10.0
A
G
B
2003 2004
2005 2006
Hog Isl. DunesHog Isl. PimplesParramore Isl.
marshshrubsummit
FIG. 16. Principal components analysis for transects based on species abundances. Axes 1-3 explained 5%, 4%, and 4%, respectively. Ordination was performed on data from all years combined; the plot was divided by year to increase clarity.
52
TABLE 8. Explanatory value of first three axes of Bray-Curtis ordination and non-metric multidimensional scaling and the ten most important species in each based on correlation (Pearson’s r).
Bray-Curtis Ordination Axis 1
Non-Metric Multidimensional Scaling Axis 1
37% r 25% r Morella cerifera 0.832 Morella cerifera 0.481 Ammophila breviligulata 0.459 Spartina patens 0.433 Panicum amarum 0.386 Scirpus pungens 0.136 Solanum carolinianum 0.093 Panicum amarum 0.061 Spartina patens 0.076 Festuca rubra 0.047 Cirsium horridulum 0.068 Parthenocissus quinquefolia 0.037 Rumex acetosella 0.066 Dichanthelium sphaerocarpon 0.027 Carex albolutescens 0.053 Carex albolutescens 0.024 Prunus serotina 0.053 Fimbristylis caroliniana 0.023 Polygonum hydropiperoides 0.036 Solanum carolinianum 0.023
Axis 2 Axis 2 7% r 21% r
Schizachyrium scoparium 0.832 Morella cerifera 0.681 Morella cerifera 0.459 Ammophila breviligulata 0.581 Aristida tuberculosa 0.386 Panicum amarum 0.412 Euphorbia ammonoides 0.093 Solanum carolinianum 0.096 Cyperus sp. 0.076 Polygonum hydropiperoides 0.087 Dichanthelium sphaerocarpon 0.068 Cirsium horridulum 0.083 Rumex acetosella 0.066 Rumex acetosella 0.068 Linaria canadensis 0.053 Prunus serotina 0.067 Eupatorium hyssopifolium 0.053 Carex albolutescens 0.055 Juncus dichotomus 0.036 Parthenocissus quinquefolia 0.047
Axis 3 Axis 3 11% r 11% r
Scirpus pungens 0.832 Baccharis halimifolia 0.230 Spartina patens 0.459 Phyla lanceolata 0.179 Ammophila breviligulata 0.386 Scirpus pungens 0.155 Morella cerifera 0.093 Juniperus virginiana 0.112 Panicum amarum 0.076 Mikania scandens 0.101 Galium sp. 0.068 Galium sp. 0.076 Phyla lanceolata 0.066 Festuca rubra 0.065 Phragmites australis 0.053 Teucrium canadense 0.059 Boehmeria cylindrica 0.053 Eupatorium capillifolium 0.058 Dichanthelium sp. 0.036 Boehmeria cylindrica 0.055
53
Diversity partitioning
Between-transect beta diversity as measured by three different indices was
significantly higher in 2006 than in other years measured (Fig. 17). When comparing beta
diversity among pimples, the three measures produced markedly different results in some
instances (e.g., compare the pimple marked with a * in Fig. 18). Conversely, some
pimples had consistently high or low beta-diversity levels (e.g., the two marked with **
and *** in Fig. 18).
2003 2004 2005 20060.0
0.5
1.0
1.5
Wils
on &
Shm
ida'
s M
easu
re
2003 2004 2005 20060
2
4
6
Cod
y 's
Mea
sure
2003 2004 2005 20060.0
0.5
1.0
1.5
Whi
ttake
r's M
easu
re
FIG. 17. Comparison of three measures of beta diversity between transects in pimples by year (Magurran 1988).
Analysis of spatial partitioning of diversity revealed different patterns for species
richness, Simpson’s diversity, and Shannon diversity (Fig. 19). Mean richness and
diversity indices in transects were significantly lower than would be expected by random
chance. The same pattern held for total annual mean richness and diversity. Partitions
determined for dune reference plots nearly exhibited the opposite pattern, having more
species per subplot than expected, and fewer per plot and dune line.
b a a a
ba a a
ca a
b
54
FIG. 18. Maps of three measures of beta diversity between transects in pimples. Asterisks are for emphasis; numbers indicate maxima and minima for each map.
Observed Expected
10
20
30
40
50
richn
ess
Observed Expected0.0
0.5
1.0
1.5
2.0
Shan
non
Div
ersi
ty In
dex
Observed Expected0.0
0.2
0.4
0.6
Sim
pson
's D
iver
sity
Inde
x
FIG. 19. Diversity partitioning within pimples as measured by three diversity indices. Expected partitions (i.e. null value estimates) are based on 10,000 resampling iterations performed with PARTITION (Crist and Veech 2006). * Observed value was either > 95% of the estimated simulations or < 5% of simulated values.
Wilson and Shmida’s
*
**
*** 0.5
1.3
Cody’s
*
**
***
2.2
8.5
Whittaker’s
*
**
*** 0.6
1.6
*
*
*
*
* *
*
*
between pimpleswithin transects between yearsbetween transects
55
TABLE 9. Diversity partitioning within and among pimples and within and among the main dunes as calculated with PARTITION.
*P-value is the proportion of null values with a diversity estimate greater than the observed. Significant probabilities are in boldface type. **Null-value estimates made from 10000 randomizations.
Species Richness Shannon Diversity Simpson’s Index Obs. Exp.** P* Obs. Exp.** P* Obs. Exp.** P* Pimples Total: (gamma) 52 1.979 0.678 Years alpha: 31 36 1 1.745 1.866 0.998 0.639 0.670 0.986 beta: 21 17 0 0.233 0.112 0.002 0.039 0.008 0.014 Pimples alpha: 9 9.4 0.545 1.247 1.23 0.152 0.579 0.5693 0.127 beta: 22 21.6 0.217 0.498 0.515 0.832 0.060 0.0695 0.871 Transects alpha: 4 4.7 1 0.754 0.848 1 0.403 0.453 1 beta: 5 5 0.508 0.493 0.424 0 0.176 0.136 0 Main dunes Total: (gamma) 39 0.8143 2.215 Years alpha: 30.3 30.3 0 0.7982 0.7982 0 2.053 2.053 0 beta: 8.7 8.7 0 0.0161 0.0161 0 0.162 0.162 0 Dune alpha: 17.1 19.8 1 0.7621 0.776 0.9833 1.806 1.899 1 ridges beta: 13.2 10.5 0 0.0361 0.0222 0.016 0.247 0.154 0 Plots alpha: 6.6 9 1 0.614 0.7076 1 1.155 1.534 1 beta: 10.5 8.1 0 0.1481 0.0545 0 0.651 0.272 0 Subplots alpha: 3.7 3.4 0 0.5593 0.5443 0 0.89 0.812 0 beta: 2.9 2.9 0.2335 0.0547 0.0758 1 0.265 0.295 1
DISCUSSION
General descriptors
Pimple-level patterns. Pimples were surprisingly species rich, considering their
small size and the relatively harsh environment on the islands (Ehrenfeld 1990). The
variability in diversity between pimples is also noteworthy. It should be considered as a
56
starting point for future study, and anyone using pimples as replicated study sites should
consider their heterogeneity.
Lack of a relationship between pimple size and different measures of diversity
was unexpected, since the species–area relationship is a basic tenet of ecology (Gleason
1925, MacArthur et al. 1966, Connor and McCoy 1979, Diamond 1988). Crist and Veech
(2006), however, have found that the relationship between area (or sampling effort) and
richness or diversity is often minimal. The close proximity of pimples to each other likely
promotes propagule transfer and may diminish the relationship (MacArthur et al. 1966).
That species richness (albeit not composition) is identical on pimple summits and dune
ridges also suggests that species dispersal and colonization of the patchy habitats on the
island is not overly influenced by patch size (Fahrig 1990, Burton and Bazzaz 1995,
Planty Tabacchi et al. 1996, Tilman 1997, Aguiar and Sala 1999).
Lack of a relationship between geographic position and diversity also suggests
that species distribution is fairly uniform across the northern end of Hog Island. When
evaluating pimples as experimental units, lack of species–area and species–location
relationships could be cited as a measure of uniformity among them. This finding does
not, however, suggest that pimples are representative of habitat conditions on the island
at large. There appears to be a bimodal distribution of species richness along the north–
south gradient, since most of the less diverse pimples were in the center of the study area
(Gauch and Chase 1974, Peet and Loucks 1977, Gauch and Stone 1979, ter Braak 1986,
Allen et al. 1991, Peltzer et al. 1998). Perhaps there is a disturbance effect from
proximity to the vehicle trail, but this is unlikely since vehicle use and even foot traffic is
very limited on the island (Hosier and Eaton 1980).
57
Transect-level patterns. The significant effect of habitat type in explaining
variation in richness and diversity indices is a good indicator that the zones are indeed
distinct. The significant drop in richness in the shrub zone of pimples is largely due to
competition from wax myrtle (Smith et al. 1995, Tolliver et al. 1995, Adler et al. 1998).
The high diversity on pimple summits indicates that the dense shrub thickets are not a
barrier to seed dispersal, even with species that are not typically wind or animal
dispersed.
Of all the dominant species on pimples and the entire island, wax myrtle, M.
cerifera, is by far the most important in terms of abundance (Young 1992, Young et al.
1995b). Most of the other important species are either marsh graminoids and forbs or
other woody and sub-woody species. This suggests either that habitats are less
heterogeneous in marsh and shrub zones relative to summit zones, i.e. fewer niches, or
that there are fewer species suited to living in those zones, a historical artifact reflecting
the suite of species that colonized the island (Diamond 1988, Houle 1997, Hofer et al.
2004).
Ordinations
The dissimilarity of species relative dominance, relative frequency, and
importance values between main dune plots and pimple summits is reflected in the
groupings of transects in the ordinations performed on the species abundance data. Xeric
plots are generally the most widely dispersed group of transects, relative to marsh and
shrub plots, and within that cloud of points, pimple summit plots are segregated from
dune plots, with the latter seeming to create a more uniform group (i.e. a tight group of
nearly-overlapping points). Marsh plots also form distinct groups, but these groups have
58
the occasional intrusion of shrub transect within them and the distinction between
reference plots and pimple plots is not as sharply defined. Shrub plots formed the tightest
groupings, meaning that transects exhibited the least variation . Pimple and reference
shrub plots overlapped frequently. The similarity of species composition in pimple and
reference shrub zones is most likely a result of light competition from wax myrtle and
subsequent reduction in species richness and abundance in the understory.
There were clear differences between pimple and dune plots, especially among
the xeric samples. This evidence suggests that pimples are not exact replicates of island
plant assemblages and may not be good models to use as experimental units. Most of the
dune reference plots are in the center of dune summits and several meters away from the
edge of the wax myrtle thickets, whereas most pimples have an open summit of only a
few meters across, many with varying levels of canopy closure. Presence of wax myrtle
and other shrub and shrub-edge species is likely a primary reason for assemblage
structure differing between dunes and pimple summits. This suggests that xeric habitat
conditions on main dunes may not be closely approximated by pimple summits.
Of the three ordination methods used, NMS and Bray-Curtis ordinations provided
markedly greater explanation of variation over PCA. This suggests that species are
responding to environmental factors in a nonmonotonic fashion (i.e., not linearly along
one resource gradient) and indicates a high level of complexity of species abundances
within pimple plant assemblages (Palmer 2007). The next chapter will explore the factors
to which species are responding.
59
Rarefaction curves
The ICE and Chao 2 estimates were greater than observed species richness
suggesting that species richness may have been underestimated in floristic studies
(Summerville and Crist 2002, Veech et al. 2002, Crist and Veech 2006). Low richness
and diversity in shrub plots reflect the density of the thicket canopy. Although M. cerifera
is a nitrogen-fixing species and deposits nitrogen-rich litter, it evidently grows too
densely to allow many species to grow underneath it (Sande and Young. 1992, Young
1992, Smith et al. 1995, Tolliver et al. 1995, Adler et al. 1998, Crawford and Young
1998a, Wijnholds and Young 2000). High diversity and richness in summit plots suggests
diverse environmental conditions or niches within those areas of pimples relative to other
zones (Gauch 1982a, Palmer 2007). The rarefaction curve of dune reference plots seems
to be closer to reaching an asymptote compared to the curve for pimple summits.
Diversity partitioning
The higher beta diversity measured in 2006 is difficult to explain. I was the only
person making field identifications, so there is little reason to believe that my personal
accuracy or bias changed dramatically in the fourth year. The differences in the 2005
survey are readily explained by smaller data set and timing of the survey (i.e., beginning
rather than height of the growing season). The 2005 season may also have been affected
by the late summer.
Diversity partitioning helped to describe the assemblage structure on pimples and
the main dunes. There was nearly equal α- and β-richness within and between transects,
which indicates that there was nearly complete turnover of species from one habitat zone
to the next (Crist et al. 2003, Crist and Veech 2006). That, in turn, is an indication of
60
successful a priori stratification of transects on the pimples. Mean annual richness for all
pimples was around three times greater than mean per-pimple richness, total richness
over the entire four years of the study was even greater, and annual variations in richness
and diversity were significantly different from random. This means that there was a
definite variation in assemblage structure among pimples and that there was a shift in
species composition over the years (Summerville and Crist 2005). Although some of this
change can be attributed to measurement error and (in the case of annual variation) an
artifact of the aberrant 2005 data set, it is still evidence of significant spatial and temporal
variation in assemblage structure on pimples.
There are significantly non-random changes in diversity between plots within
each of the three main dune lines and between the dune lines themselves. The latter
finding can be explained by the differing ages of the dunes and presumably different
successional stages existing in each. The former is explainable by the heterogeneity of
placement of the plots within each ridge. There was no significant difference between
observed annual changes and those expected at random. This implies that spatial
variation in main dunes is similar to that in pimples but temporally, main dune
assemblages are more stable (Summerville and Crist 2005).
CONCLUSIONS
At the beginning of this study, the zonation of plant assemblages on pimples and
the main dunes appeared similar. Although pimples apparently have similar species
richness and diversity patterns as the dunes near them, the actual assemblage composition
appears to differ. Pimples also seem to be more prone to shifts in species assemblage than
61
do the main dunes. Pimple summit communities are the most species-diverse and
putatively most diverse relative to the other zones. Summits also have apparently more
edge effect than the wider areas of dune xeric patches and more overlap in species
composition with shrub and marsh assemblages. All of these factors should be seen as
caveats if using pimples as experimental units. Species assemblages on pimples do not
exactly duplicate those on main dunes.
These differences in species composition between pimples and main dunes are
likely related to the ways in which each form. Whether biology drives geology during the
formation of pimples or the reverse remains to be determined. In the case of tundra tree
islands there apparently has to be a tree colonizing first before , although abiotic factors
may influence establishment (Thomson 1950, Meredith 1972, Marr 1977, Davis 1980,
Benedict 1984, Holtmeier and Broll 1992, Pauker and Seastedt 1996, Parker and Sanford
1999, Seastedt and Adams 2001). With tree islands in swamps, a geologic formation must
be in place before the community can become established (Svihla 1930, 1939, Loveless
1959, Rich and Spackman 1979, Duever and Riopelle 1983, Troxler Gann et al. 2005).
Future studies comparing pimples of different ages may resolve this question.
The newer or less commonly used mathematical methods employed for this study
show promise for future fieldwork. When comparing statistical methods used in this
study, ordinations have some definite advantages over general linear models (ANOVAs).
There is no need for a priori construction of models, no concerns over loss of degrees of
freedom, and more robust handling of data sets with many transects and species (Palmer,
2007). Moreover, the graphical representation of variations and associations between
transects (or species) can make interpreting results more intuitive, albeit more artful.
62
Although it would be ideal to have a prior idea of the nature of variance in species
responses and to choose an ordination technique accordingly, a posteriori comparison of
ordination results provided valuable insight into the nature of variation in species
responses, namely that species are responding in a nonmonotonic fashion to
environmental gradients. Partition estimation of diversity has the decided advantage over
the formulaic measures of calculating β-diversity, since it provides measures of diversity
in the same units at all levels. The more standard measures of β-diversity only allowed
for cross comparison within one level.
63
CHAPTER 4
ENVIRONMENTAL INFLUENCES ON SPECIES DISTRIBUTION
INTRODUCTION
A basic research goal of plant ecology is to understand the influences of
environment on assemblage structure (Hayden et al. 1995). This is important and
fundamental for understanding assemblage-level, community-level, and ecosystem-level
functioning and has practical uses for ecosystem monitoring and restoration. The plant
assemblages and ecosystems of barrier islands and dunes were among the first to be
studied by ecologists, e.g., (Cowles 1899, Kearney 1904). Olsson-Seffer wrote in 1909
that “In discussing the factors that influence plant life [on dunes], I have found it
convenient to classify them into the following groups: atmospheric, hydrodynamic,
edaphic, topographic, and historical factors.” A century later, much remains to be
understood about their ecology (Ehrenfeld 1990).
The readily apparent ecological driver on the islands is water. The position of the
rain-charged fresh water table relative to the soil surface is the primary determinant of
community type in the interior of barrier islands at the VCR. The difference between
freshwater marsh, shrub or bramble thickets, and xeric dunes is apparently the result of
distance above the fresh water table (Hayden et al. 1991, Hayden et al. 1995).
‘Elevational’ ranges containing those communities are narrow, on the scale of
decimeters. Despite its apparent importance, availability of fresh water can only explain
gross differences between communities and their inherent plant assemblages. In that case,
species composition in assemblages should be largely uniform at equal elevations on each
island. I have not observed that to be true.
64
There are a few approaches to describing and defining the role of environment in
determining species composition. Experimental tactics would be to use reciprocal
transplants, glass house experiments under different conditions, and experimental
modification of assemblage composition, i.e., removal or addition of species (Tilman
1997). Alternatively, descriptive studies relating species abundances to environmental
variables can be used in multivariate analyses and ordinations (Peet and Loucks 1977,
Gauch 1982a, Pielou 1984, Peet et al. 1988, Palmer 2007).
I hypothesize that there are synergistic effects between water table, soil,
landscape, and biota that elicit variation in plant species abundances within each
community zone. This is not a novel idea in plant ecology (Olsson-Seffer 1909, Curtis
and McIntosh 1951, Peet and Loucks 1977, Gauch 1982a, Pielou 1984, Peet et al. 1988,
Frego and Carleton 1995, Bazzaz 1996, Palmer 2007), but there is nothing approaching a
unified theory of plant species–environment interaction as yet. I studied relationships
between plant assemblage composition and microhabitat conditions to determine what
drivers were creating such tightly-packed habitat zones and to see if there were
differences in species – environment interactions between pimples and main dunes.
MATERIALS AND METHODS
The study area and floristic survey methodology are the same as described in
Chapters 2 and 3. Besides taking the plant censuses, I also collected environmental data
and analyzed them as described below.
Environmental measurements: hydrology
To determine depth to water table at a transect, I used a 10 cm–diameter soil
auger to dig down to the ground water. To find the distance, I either used a weighted
65
water sensor whose LED lit upon contact with water, or, for depths less than 30 cm, a
ruler. Marsh soils on the barrier islands often have a thick muck layer, and those under
shrubs a tight duff layer. To account for this difference, I measured water table depth as
the depth from the top of the mineral horizon. Over two days in summer 2004, I used a
portable conductivity meter to test water samples taken from bore holes for salinity and
pH. For the next two years, I used soil salinity and pH data as determined by the soil
testing lab at Virginia Polytechnic Institute and State University (VPI) (Mullins and
Heckendorn 2006).
In addition to sampling as described above during vegetation surveys, during
summers, I periodically checked water levels at as many pimples as possible, as dictated
by rain and drought events. I did this in combination with inspection of water table data
from automated water table wells installed on the island in order to get a reasonable
measure of maximum and minimum water levels during the study.
Environmental measurements: edaphic parameters
To sample surface soil for a transect, I used a soil corer made from a plastic
syringe with the end cut off to take three 15 cm3 cores at the ends and middle of the
transect. The three cores were mixed together for analysis to produce an average
composite soil sample for the transect and stored in Whirl-Pak™ sterile sample bags
(Nasco Co.). I used a soil auger to sample soil at depths determined by examining the soil
profile for changes in color, an indication of redox potential and water table position
(Brown et al. 1990, Huggett 1998, Silver et al. 1999, Brady and Weil 2002). Hayden et
al. (1995) found that the soil profile under Parramore Island pimples was a homogeneous
sand horizon. After performing a particle size analysis on twenty samples from different
66
depths in the dunes, I found that, regardless of depth, the soil profile in Hog Island
pimples was composed of a well-sorted fine sand. A mean 92% of each sample was in the
particle size range of 2-3 φ (125 – 250 µm diam.). Particle size therefore was not useful
in discriminating microhabitats, and I did not use it in analyses.
Both the depth of the organic horizon and the amount of organic matter in the
mineral horizon of the soil were variable, however. These two factors are important in
sandy soils for water and nutrient retention, and were therefore recorded. Soil organic
matter was determined by mass loss on ignition. Depth of the organic layer was
determined by observing excavated bore holes; there was typically a distinct interface
between organic matter and sand.
During spring and summer 2005 and summer 2006, I collected samples for
chemical analysis at the VPI soil testing laboratory. The routine battery of tests from the
lab assessed levels of P, K, Ca, Mg, Zn, Mn, Cu, Fe, B, cation exchange capacity (CEC),
Ca saturation (CaSat), Mg saturation (MgSat), K saturation (KSat), organic matter in the
mineral (A) horizon from mass lost on ignition (OM), pH and salinity (Mullins and
Heckendorn 2006). During the same collections in summer 2005 and 2006 and during
spring 2006, I collected samples for NH4+ (NH4) and NO2
- – NO3- (NOx) analysis, but the
2005 sample could not be used. Within the day I collected them, I extracted the samples
with a 2N KCl solution and froze them in Whirl-Pak bags to await analysis. I determined
NH4+ and NOx
levels colorimetrically on a Lachat analyzer in the Department of
Environmental Sciences at the University of Virginia. Subsamples of soil collected for N
analysis were air and subsequently oven dried to determine gravimetric water content.
67
These measurements served both as environmental parameters and as standards for KCl-
extracted samples to determine actual ion content in soil.
Statistical analyses
One ordination method expressly designed to relate assemblage composition to
environmental factors is canonical correlation analysis or CCA (Kent and Ballard 1988,
Kourtev et al. 1998). This multivariate ordination technique uses a species abundance
matrix to determine multidimensional correlations among species. Mutually orthogonal
axes (usually two or three) are created that describe a percentage of variation either
among transects or in species’ abundances, and vectors representing environmental
factors are overlaid on them (Gauch 1982a, 1982b, Kent and Ballard 1988). The ‘spatial’
relationship between transects or species with environmental factors in ordination space
represents the influence of those factors. Assumptions of CCA are multivariate normality,
linear relationships between variables (i.e., CCA is a multiple multiple regression), and
orthogonality of environmental factors (Gauch 1982a, Økland 1996). Multivariate
normality is difficult to evaluate and achieve (ter Braak 1986, Palmer 2007), but CCA is
robust to less than normal data (Minchin 1987, McCune and Mefford 1999). Monte-Carlo
tests can determine if axes significantly describe linear relationships within the data
matrices (McCune and Mefford 1999). A presumption of CCA is that environmental
factors used are meaningful and appropriate to the community being studied; it is
therefore valid practice to remove unimportant or highly-correlated environmental
variables and perform analyses again to improve the solution (Økland 1996). One
limitation of CCA is that calculation of its distance matrix (ordination solution) can
68
exaggerate contributions from rare species (Faith et al. 1987, Minchin 1987, Økland
1996).
Alternatively, multiple ANOVA’s, MANOVA’s, or regressions could be used to
provide a complimentary analysis of the data. Although they may be helpful in a
posteriori evaluation of single species responses, ANOVA’s and MANOVA’s have
neither robustness against data sets with many zeroes nor the ability to evaluate
relationships between all species and environmental gradients simultaneously (ter Braak
1986, McCune and Mefford 1999). Single or stepwise linear regressions are similarly
insufficient compared to multivariate, multidimensional approaches (Pausas and Austin
2001).This makes CCA a more appropriate test for examining all factors simultaneously.
(ANOVA analyses of habitat and location differences in edaphic factors are nevertheless
included here.)
Although canonical correspondence analysis does have the attractiveness of being
a direct-gradient test as opposed to ordinations based on pure statistical distance, it
explains less variation in a data matrix than indirect-gradient ordinations (Minchin 1987,
Kent and Ballard 1988, Økland 1996, McCune and Mefford 1999, Palmer 2007). In
practice, cumulative percent variation described by the first three axes of a CCA
ordination are often in the range of 20 % – 40 % (2001). In contrast, an indirect method
like non-metric multidimensional scaling, which is optimized either manually or with an
‘auto pilot’ algorithm, often explains > 80 % of data variation in the first 1–3 axes
(McCune and Mefford 1999, Palmer 2007, personal observation). This should not detract
from using CCA, however. If one assumes that CCA is directly creating regression
69
relationships between environmental variables and species abundances, explaining 20 %
– 40 % of the drivers of diversity in a community is respectable.
Ordinations. Because the data set is not complete for all years, all transects, and
all environmental factors, I could not inspect all data together in one ordination. As such,
I ran CCA analyses in two different configurations: 1) with annual mean values from all
transects and a constrained set of environmental variables, and 2) with annual mean
values from only pimple transects and all environmental characteristics that I recorded.
I also performed analyses with those two configurations but with matrices
transposed so that I could examine relationships of species to transects and, more
importantly, to environmental variables. To create a species – environmental factors
matrix, I calculated mean values of each environmental variable from all transects in
which a particular species occurred.
Graphical results from these ordinations will have species labeled with my
arbitrary, a priori habit preference: xerophytic, mesophytic, or hydrophytic. I attempted
to base my suppositions not only on the island habitats in which I routinely observed each
species, but also on the drought tolerances and water needs those species have on the
mainland. For example, I classified little blue stem, Schizachyrium scoparium, as
mesophytic because it is typically a prairie species, even though I only found it in some
of the driest dune sites on Hog Island.
In addition to CCA ordinations, I calculated non-metric multidimensional scaling
analyses of both transects as well as species based on environmental variables. I
performed each analysis in the two configurations described above. I wanted to compare
performance of a direct-gradient approach (CCA) to an indirect approach (NMS).
70
RESULTS
Comparison of environmental factors by habitat type
Many environmental factors were significantly different between habitat types.
There was often a difference in soil mineral content between marsh and the other
habitats; this could be attributed to loss by leaching in shrub and summit zones (Fig. 20).
Possibly because there was so little nitrogen in general, there was no significant pattern in
nitrogen distribution, except for significantly higher nitrate–nitrite levels in main dunes
versus pimples (Fig. 21). Depth of water table differed between habitat zones as well as
between pimple and main dune (Fig. 21).
Since some of the distribution patterns of edaphic factors were similar, I
calculated Pearson’s correlation coefficients for all environmental factors (McCune and
Mefford 1999). These coefficients are on a scale from 1 to -1 and measure the degree of
linear relationship between variables. Boron, calcium, and magnesium were highly
correlated with each other and with cation exchange capacity, but not with salinity.
Copper, zinc, and phosphorus were all correlated with each other 50–70 %. Many of
those minerals had either a positive correlation with water table depth that was greater
than 50 %, a negative correlation with elevation that was less than -50 %, or both. Iron
was correlated with organic matter content. Ammonium and nitrate–nitrite were not
highly correlated with any factor, including themselves.
There were fewer correlations between physical features and other factors.
Elevation, depth to water table, and organic layer thickness were correlated Water and
organic horizon thickness were correlated with some mineral nutrients and cation
exchange capacity.
71
FIG. 20. Content of several nutrients, base saturations, organic matter content, and cation exchange capacity for marsh, shrub, and xeric habitat soils on pimples and main dunes. Data were analyzed with a one-way, full-factor ANOVA; letters indicate significant differences between habitats found with Tukey’s multiple comparison test at α = 0.05. Significant pimple–main dune effects are indicated with * (p < 0.05), ** (p < 0.01), and *** ( p < 0.001). KEY: marsh, pimple; marsh, main dune; shrub, pimple; shrub, main dune; summit, pimple; summit, main dune; pimple total; main dune total.
0.5
1.0
1.5 a
b
c
**
boro
n (p
pm)
250
500
750
1000
calc
ium
(ppm
)
a a
b
***
25
50
75
% C
a ba
se s
atur
atio
n
a ab b b *
2
4
6
8
CEC
(meq
/ 10
0 g)
a
b
c
***
0.05
0.10
0.15
0.20
copp
er (p
pm)
a
b b
50
100
150
iron
(ppm
) a b
a
***
100
200
300
400
pota
ssiu
m (p
pm)
10
20
30
% K
bas
e sa
tura
tion
100
200
300
400
mag
nesi
um (p
pm)
a
b
c
**
20
40
60
% M
g ba
se s
atur
atio
n a a b **
1
2
3
4
5
man
gane
se (p
pm)
2
4
6
8
% o
rgan
ic m
atte
r
ab a
b
*
72
FIG. 21. Ammonium, nitrate–nitrite, phosphorus, and zinc; organic horizon thickness; salinity; % slope, and water table for three different habitats on pimples and main dunes. Data for organic horizon thickness, salinity (conductivity), and slope were only available for pimples in amounts sufficient for analysis. Data were analyzed with a one-way, full-factor ANOVA; letters indicate significant differences between habitats found with Tukey’s multiple comparison test at α = 0.05. Significant pimple–main dune effects are indicated with * (p < 0.05), ** (p < 0.01), and *** ( p < 0.001). KEY: marsh, pimple;
marsh, main dune; shrub, pimple; shrub, main dune; summit, pimple; summit, main dune; pimple total; main dune total.
wat
er ta
ble
posi
tion
(cm
)
-150
-100
-50
0
100
200
300
400
amm
oniu
m (p
pm)
100
200
300
400
500
*
nitr
ate
& n
itrite
(ppm
)
10
20
30
phos
phor
us (p
pm)
* b b
a
0.50
1.00
1.50
2.00
Zinc
(ppm
) a b
b
5
10
15
O-h
oriz
on th
ickn
ess
(cm
)
a
b
a
500
1000
1500
salin
ity (p
pm)
2
4
6
8
% s
lope
a b b
73
TABLE 10. Pearson’s correlation coefficients for environmental factors used in the study. ‘Elevation’ refers to height above marsh. ‘O horizon’ is the thickness of the organic horizon of the soil. ‘East’ and ‘north’ refer to degree of aspect east to west or north to south, respectively. ‘Salinity’ was measured as soil water conductivity. ‘Water’ is depth to the water table. ‘CEC’ is cation exchange capacity. The ‘Sat’ suffix indicates percent base saturation. ‘NOx’ is nitrate–nitrite concentration.
(part 1) O hor. east north salinity slope water P K Ca Mg Zn
elevation -0.71 -0.08 0.05 -0.28 0.32 -0.89 -0.38 -0.12 -0.54 -0.52 -0.50
O horizon -0.01 0.14 0.25 0.01 0.75 0.22 0.23 0.54 0.59 0.26
east 0.09 -0.04 0.06 0.03 -0.21 -0.05 0.02 0.00 0.02
north -0.03 0.13 -0.02 -0.03 -0.03 -0.11 -0.17 0.04
salinity -0.08 0.36 0.00 0.04 0.04 0.26 0.19
slope -0.35 -0.40 0.06 -0.19 -0.07 -0.49
water 0.41 0.12 0.65 0.65 0.56
P -0.04 0.61 0.29 0.70
K 0.03 0.17 -0.08
Ca 0.83 0.60
Mg 0.38
(part 2) Mn Cu Fe B CEC CaSat MgSat KSat OM NH4 NOx
elevation 0.27 -0.38 -0.04 -0.57 -0.55 0.05 -0.08 0.03 0.06 0.07 0.26
O horizon -0.11 0.37 0.12 0.52 0.61 -0.17 0.12 0.08 0.09 0.10 -0.32
east 0.32 -0.10 0.06 -0.07 -0.01 0.05 -0.06 0.01 -0.02 -0.04 -0.04
north 0.06 -0.07 0.07 -0.10 -0.14 0.05 -0.08 0.03 -0.02 -0.10 0.00
salinity -0.24 -0.03 0.45 0.19 0.15 -0.37 0.42 -0.03 0.15 -0.05 -0.03
slope 0.07 -0.45 0.28 -0.23 -0.11 -0.15 0.09 0.10 0.23 0.17 0.17
water -0.19 0.45 0.10 0.67 0.66 -0.07 0.11 -0.04 0.02 0.09 -0.33
P 0.19 0.68 -0.49 0.61 0.43 0.52 -0.50 -0.10 -0.56 -0.06 -0.31
K -0.08 -0.03 -0.11 0.09 0.44 -0.48 -0.22 0.95 0.14 -0.13 -0.12
Ca 0.30 0.49 -0.21 0.91 0.88 0.36 -0.28 -0.14 -0.14 0.36 -0.36
Mg 0.05 0.34 0.14 0.87 0.92 -0.15 0.18 -0.02 0.26 0.43 -0.21
Zn 0.13 0.61 -0.10 0.66 0.44 0.29 -0.20 -0.15 -0.22 -0.09 -0.32
Mn -0.05 -0.32 0.09 0.13 0.55 -0.55 -0.07 -0.27 0.19 -0.07
Cu -0.25 0.53 0.39 0.22 -0.15 -0.11 -0.24 -0.14 -0.25
Fe -0.07 -0.08 -0.56 0.73 -0.15 0.74 0.06 0.16
B 0.87 0.14 -0.10 -0.07 -0.06 0.39 -0.25
CEC -0.07 -0.13 0.26 0.10 0.32 -0.31
CaSat -0.73 -0.47 -0.65 0.11 -0.18
MgSat -0.26 0.64 -0.01 0.24
KSat 0.09 -0.15 -0.05
OM 0.20 0.09
NH4 0.19
74
Transect ordinations: combined main dune and pimple data
Overall, total variation explained in the first three (or two) axes of non-metric
multidimensional scaling ordinations was at least three times as much as in the first three
axes produced by canonical correspondence analysis ordinations. Nevertheless, in every
CCA ordination, all axes significantly described linear relationships between variables,
i.e., the relationship between factor and transect matrices or factor and species matrices
was not random (Monte-Carlo, P < 0.05 for all). In every NMS ordination performed,
resulting axes explained variation significantly better than would be expected by chance
(Monte-Carlo test of ‘stress’ reduction; P < 0.02 for all axes).
Canonical correspondence analysis of transects produced a solution whose first
three axes together described 21.1 % of the relationship between assemblage data and
environmental variables (Table 11). The ordination revealed some distinct groupings of
sites based on habitat type (Fig. 22): Summit plots appeared most variable. Main dune
plots were clustered more tightly than pimple plots. Habitat types sorted themselves
along axis 1, which was correlated with water table depth (Fig. 23) and mineral
micronutrients (Fig. 24–Fig. 27). Summit and marsh transects together segregated from
shrub transects along axis 2, which was correlated with phosphorus (Fig. 28; low in shrub
transects, high elsewhere) and iron (Fig. 27; high in shrubs, low elsewhere). Variation
among transects within each habitat along axis 3 was correlated with nitrate–nitrite levels
(Fig. 29), cation-exchange capacity (Fig. 30), and organic matter (Fig. 31).
75
TABLE 11. Pearson’s r correlation coefficients of environmental variables for the first three axes of a CCA of main dunes and pimples and percentage of variation explained. In all subsequent tables, percentages listed with axes refer to the percent of variation explained by that particular axis.
Axis 1: 9 % Axis 2: 7 % Axis 3: 5 % R r r water table 0.954 P -0.675 NOx 0.887 B 0.601 Fe 0.586 CEC 0.638 Zn 0.572 Zn -0.543 OM 0.616 Mg 0.535 MgSat 0.53 Ca 0.578 Fe 0.502 B -0.44 Mg 0.551 MgSat 0.497 Cu -0.436 K 0.452 CEC 0.423 CaSat -0.431 B 0.385 Mn -0.377 OM 0.415 MgSat -0.348 Ca 0.371 Mn -0.312 Zn 0.32 Cu 0.336 NOx 0.213 Mn 0.204 P 0.316 Ca -0.121 CaSat 0.199 CaSat -0.296 CEC -0.072 NH4 0.184 OM 0.242 KSat -0.069 Fe -0.153 KSat -0.184 K 0.052 KSat 0.137 NOx -0.122 water table 0.038 P 0.097 NH4 -0.051 Mg -0.027 Cu 0.091 K 0.025 NH4 0.011 water table 0.01
1
-2
-1
0
1
2
-2-1
01 -2
-10
G
A
B
A
B
G
A
B
G
B
GA
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
BG
A
B
B
G
A
B
BB
B
B
B
BB
G
A
B
B
G
A
B
B
B
G
A
A
A
G
B
PIMPLESMAIN DUNES
MARSHSHRUBSUMMIT
CC
A A
XIS
2: 7
%
CCA AXIS 1: 9%
CCA AXIS 3: 5%
watertable
P
K
CaMg
Zn
Mn
Cu
Fe
B
CEC
CaSat
MgSat
KSat
OM
NH4
NO3
1
-2
-1
0
1
2
-2-1
01 -2
-10
G
A
B
A
B
G
A
B
G
B
GA
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
BG
A
B
B
G
A
B
BB
B
B
B
BB
G
A
B
B
G
A
B
B
B
G
A
A
A
G
B
PIMPLESMAIN DUNES
MARSHSHRUBSUMMIT
CC
A A
XIS
2: 7
%
CCA AXIS 1: 9%
CCA AXIS 3: 5%
watertable
P
K
CaMg
Zn
Mn
Cu
Fe
B
CEC
CaSat
MgSat
KSat
OM
NH4
NO3
FIG. 22. Canonical correspondence analysis ordination of main dune and pimple transects. In this figure and all other CCA figures like it, lengths of environmental factor vectors have been exaggerated by a factor of two for legibility. In all other figures, percentages listed on axes refer to the percentage of variation explained.
76
CC
A A
XIS
2: 7
%
CCA AXIS 1: 9%
CCA AXIS 3: 5%
1
-2
-1
0
1
2
-2-1
01 -2
-10
G
A
B
A
B
G
A
B
G
BGA
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
BG
A
B
B
G
A
B
BB
B
B
B
BB
G
A
B
B
G
A
B
B
B
G
A
A
A
G
B
PIMPLESMAIN DUNES
MARSHSHRUBSUMMIT
-116 -45 25
A
A
A
water table position (cm)
CC
A A
XIS
2: 7
%
CCA AXIS 1: 9%
CCA AXIS 3: 5%
CC
A A
XIS
2: 7
%
CCA AXIS 1: 9%
CCA AXIS 3: 5%
1
-2
-1
0
1
2
-2-1
01 -2
-10
G
A
B
A
B
G
A
B
G
BGA
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
BG
A
B
B
G
A
B
BB
B
B
B
BB
G
A
B
B
G
A
B
B
B
G
A
A
A
G
B
PIMPLESMAIN DUNES
MARSHSHRUBSUMMIT
-116 -45 25
A
A
A
water table position (cm)
FIG. 23. Overlay of water table position on the CCA ordination of pimple and main dune transects.
77
1
-2
-1
0
1
2
-2-1
01 -2
-10
G
A
B
A
B
G
A
B
G
B
GA
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
B
G
A
B
B
B
B
B
B
BB
G
A
B
B
G
A
B
B
B
G
A
A
A
G
B
PIMPLESMAIN DUNES
MARSHSHRUBSUMMIT
0.10 0.78 1.45
A
A
A
soil boron (ppm)
CC
A A
XIS
2: 7
%
CCA AXIS 1: 9%
CCA AXIS 3: 5%1
-2
-1
0
1
2
-2-1
01 -2
-10
G
A
B
A
B
G
A
B
G
B
GA
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
B
G
A
B
B
B
B
B
B
BB
G
A
B
B
G
A
B
B
B
G
A
A
A
G
B
PIMPLESMAIN DUNES
MARSHSHRUBSUMMIT
0.10 0.78 1.45
A
A
A
soil boron (ppm)
CC
A A
XIS
2: 7
%
CCA AXIS 1: 9%
CCA AXIS 3: 5%
CC
A A
XIS
2: 7
%
CCA AXIS 1: 9%
CCA AXIS 3: 5%
FIG. 24. Overlay of soil boron concentration on the CCA ordination of pimple and main dune transects.
78
1
-2
-1
0
1
2
-2-1
01 -2
-10
G
A
B
A
B
G
A
B
G
B
GA
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
BG
A
B
B
G
A
B
BB
B
B
B
BB
G
A
B
B
G
A
B
B
B
G
A
A
A
G
B
PIMPLESMAIN DUNES
MARSHSHRUBSUMMIT
0.40 1.40 2.40
A
A
A
soil zinc (ppm)
CC
A A
XIS
2: 7
%
CCA AXIS 1: 9%
CCA AXIS 3: 5%1
-2
-1
0
1
2
-2-1
01 -2
-10
G
A
B
A
B
G
A
B
G
B
GA
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
BG
A
B
B
G
A
B
BB
B
B
B
BB
G
A
B
B
G
A
B
B
B
G
A
A
A
G
B
PIMPLESMAIN DUNES
MARSHSHRUBSUMMIT
0.40 1.40 2.40
A
A
A
soil zinc (ppm)
CC
A A
XIS
2: 7
%
CCA AXIS 1: 9%
CCA AXIS 3: 5%
CC
A A
XIS
2: 7
%
CCA AXIS 1: 9%
CCA AXIS 3: 5%
FIG. 25. Overlay of soil zinc concentration n on the CCA ordination of pimple and main dune transects.
79
CC
A A
XIS
2: 7
%
CCA AXIS 1: 9%
CCA AXIS 3: 5%
1
-2
-1
0
1
2
-2-1
0 1 -2-1
0
G
A
B
A
B
G
A
B
G
B
GA
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
B
G
A
B
B
B
B
B
B
BB
G
A
B
B
G
A
B
B
B
G
A
A
A
G
B
PIMPLESMAIN DUNES
MARSHSHRUBSUMMIT
16 224 431
A
A
A
soil magnesium(ppm)
CC
A A
XIS
2: 7
%
CCA AXIS 1: 9%
CCA AXIS 3: 5%
1
-2
-1
0
1
2
-2-1
0 1 -2-1
0
G
A
B
A
B
G
A
B
G
B
GA
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
B
G
A
B
B
B
B
B
B
BB
G
A
B
B
G
A
B
B
B
G
A
A
A
G
B
PIMPLESMAIN DUNES
MARSHSHRUBSUMMIT
16 224 431
A
A
A
soil magnesium(ppm)
FIG. 26. Overlay of soil magnesium concentration on the CCA ordination of pimple and main dune transects.
80
CC
A A
XIS
2: 7
%
CCA AXIS 1: 9%
CCA AXIS 3: 5%1
-2
-1
0
1
2
-2-1
01 -2
-10
G
A
B
A
B
G
A
B
G
B
GA
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
BG
A
B
B
G
A
B
BB
B
B
B
BB
G
A
B
B
G
A
B
B
B
G
A
A
A
G
B
PIMPLESMAIN DUNES
MARSHSHRUBSUMMIT
28 129 229
A
A
A
soil iron (ppm)
CC
A A
XIS
2: 7
%
CCA AXIS 1: 9%
CCA AXIS 3: 5%
CC
A A
XIS
2: 7
%
CCA AXIS 1: 9%
CCA AXIS 3: 5%1
-2
-1
0
1
2
-2-1
01 -2
-10
G
A
B
A
B
G
A
B
G
B
GA
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
BG
A
B
B
G
A
B
BB
B
B
B
BB
G
A
B
B
G
A
B
B
B
G
A
A
A
G
B
PIMPLESMAIN DUNES
MARSHSHRUBSUMMIT
28 129 229
A
A
A
soil iron (ppm)
FIG. 27. Overlay of soil iron concentration on the CCA ordination of pimple and main dune transects.
81
1
-2
-1
0
1
2
-2-1
01 -2
-10
G
A
B
A
B
G
A
B
G
B
GA
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B G
A
B
B
G
A
B
BB
B
B
B
BB
G
A
B
B
G
A
B
B
B
G
A
A
A
G
B
PIMPLESMAIN DUNES
MARSHSHRUBSUMMIT
5.00 19.75 34.50
A
A
A
soil phosphorus(ppm)
CC
A A
XIS
2: 7
%
CCA AXIS 1: 9%
CCA AXIS 3: 5%1
-2
-1
0
1
2
-2-1
01 -2
-10
G
A
B
A
B
G
A
B
G
B
GA
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B G
A
B
B
G
A
B
BB
B
B
B
BB
G
A
B
B
G
A
B
B
B
G
A
A
A
G
B
PIMPLESMAIN DUNES
MARSHSHRUBSUMMIT
5.00 19.75 34.50
A
A
A
soil phosphorus(ppm)
CC
A A
XIS
2: 7
%
CCA AXIS 1: 9%
CCA AXIS 3: 5%
CC
A A
XIS
2: 7
%
CCA AXIS 1: 9%
CCA AXIS 3: 5%
FIG. 28. Overlay of soil phosphorus concentration on the CCA ordination of pimple and main dune transects.
82
1
-2
-1
0
1
2
-2-1
01 -2
-10
G
A
B
A
B
G
A
B
G
B
G
A
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
BG
A
B
B
G
A
B
B BB
B
B
BB
G
A
B
B
G
A
B
B
B
G
A
A
A
G
B
PIMPLESMAIN DUNES
MARSHSHRUBSUMMIT
0 121 242
A
A
A
soil NOx (ppt)
1
-2
-1
0
1
2
-2-1
01 -2
-10
G
A
B
A
B
G
A
B
G
B
G
A
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
BG
A
B
B
G
A
B
B BB
B
B
BB
G
A
B
B
G
A
B
B
B
G
A
A
A
G
B
PIMPLESMAIN DUNES
MARSHSHRUBSUMMIT
0 121 242
A
A
A
soil NOx (ppt)
FIG. 29 Overlay of soil nitrate – nitrite concentration on the CCA ordination of pimple and main dune transects.
83
1
-2
-1
0
1
2
-2-1
01 -2
-10
G
A
B
A
B
G
A
B
G
B
GA
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
B
G
A
B
B
B
B
B
B
BB
G
A
B
B
G
A
B
B
B
G
A
A
A
G
B
PIMPLESMAIN DUNES
MARSHSHRUBSUMMIT
0.40 4.63 8.85
A
A
A
cation-exchange capacity(meq /100 g)
CC
A A
XIS
2: 7
%
CCA AXIS 1: 9%
CCA AXIS 3: 5%1
-2
-1
0
1
2
-2-1
01 -2
-10
G
A
B
A
B
G
A
B
G
B
GA
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
B
G
A
B
B
B
B
B
B
BB
G
A
B
B
G
A
B
B
B
G
A
A
A
G
B
PIMPLESMAIN DUNES
MARSHSHRUBSUMMIT
0.40 4.63 8.85
A
A
A
cation-exchange capacity(meq /100 g)
1
-2
-1
0
1
2
-2-1
01 -2
-10
G
A
B
A
B
G
A
B
G
B
GA
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
B
G
A
B
B
B
B
B
B
BB
G
A
B
B
G
A
B
B
B
G
A
A
A
G
B
PIMPLESMAIN DUNES
MARSHSHRUBSUMMIT
0.40 4.63 8.85
A
A
A
cation-exchange capacity(meq /100 g)
CC
A A
XIS
2: 7
%
CCA AXIS 1: 9%
CCA AXIS 3: 5%
CC
A A
XIS
2: 7
%
CCA AXIS 1: 9%
CCA AXIS 3: 5%
FIG. 30. Overlay of soil cation-exchange capacity on the CCA ordination of pimple and main dune transects.
84
1
-2
-1
0
1
2
-2-1
0 1 -2-1
0
G
A
B
A
B
G
A
B
G
B
GA
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
B
G
A
B
B
B
B
B
B
BB
G
A
B
B
G
A
B
B
B
G
A
A
A
G
B
PIMPLESMAIN DUNES
MARSHSHRUBSUMMIT
0.3 3.2 6.1
A
A
A
soil organic matter (%)
CC
A A
XIS
2: 7
%
CCA AXIS 1: 9%
CCA AXIS 3: 5%
1
-2
-1
0
1
2
-2-1
0 1 -2-1
0
G
A
B
A
B
G
A
B
G
B
GA
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
B
G
A
B
B
B
B
B
B
BB
G
A
B
B
G
A
B
B
B
G
A
A
A
G
B
PIMPLESMAIN DUNES
MARSHSHRUBSUMMIT
0.3 3.2 6.1
A
A
A
soil organic matter (%)
1
-2
-1
0
1
2
-2-1
0 1 -2-1
0
G
A
B
A
B
G
A
B
G
B
GA
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
B
G
A
B
B
B
B
B
B
BB
G
A
B
B
G
A
B
B
B
G
A
A
A
G
B
PIMPLESMAIN DUNES
MARSHSHRUBSUMMIT
0.3 3.2 6.1
A
A
A
soil organic matter (%)
CC
A A
XIS
2: 7
%
CCA AXIS 1: 9%
CCA AXIS 3: 5%
CC
A A
XIS
2: 7
%
CCA AXIS 1: 9%
CCA AXIS 3: 5%
FIG. 31. Overlay of soil organic matter concentration on the CCA ordination of pimple and main dune transects.
85
Analysis of explanatory power changes between subsequent axes in non-metric
multidimensional scaling revealed that a three-dimensional solution was optimal. The
resulting three axes in the solution described 55 %, 21 %, and 20% of the variation (96 %
cumulative; Table 12; Fig. 32). The two factors most highly correlated to each axis were,
respectively, potassium and potassium saturation (Fig. 33), water and ammonium (Fig.
34, Fig. 35), and calcium and cation-exchange capacity.
TABLE 12. Pearson’s r correlation coefficients for the three axes of a NMS solution of main dune and pimple transects versus environmental factors.
Axis 1: 55 % Axis 2: 21 % Axis 3: 20 % r r r KSat -0.674 water table -0.608 Ca 0.736 K -0.612 NH4 0.477 CEC 0.693 NH4 -0.543 CEC -0.45 MgSat -0.608 CaSat 0.48 K -0.448 B 0.592 NOx -0.344 Mg -0.433 Mg 0.583 Cu 0.216 CaSat 0.418 CaSat 0.512 P 0.214 Zn -0.406 Mn 0.497 OM -0.15 B -0.402 Fe -0.475 Zn 0.145 Ca -0.33 NH4 0.433 MgSat 0.143 Mn 0.299 Zn 0.403 Mn -0.105 Fe -0.297 NOx 0.337 water table 0.087 KSat -0.274 P 0.326 Ca 0.091 Cu -0.259 OM 0.275 CEC -0.084 OM -0.244 K 0.232 Fe -0.045 P -0.216 Cu 0.182 B 0.047 MgSat -0.18 KSat 0.061 Mg -0.013 NOx 0.026 water table -0.014
Summary. CCA and NMS found different patterns in the transects. CCA, based on
linear relationships between factors and transects found water availability and cationic
minerals to be the most important factors for discriminating groups of transects. NMS ,
which finds the optimal solution for describing differences between transects focused
more on potassium and ammonium but otherwise found similar groupings to CCA.
86
-1
0
1N
MS
AXI
S 2:
21%
-2-1
0NMS AXIS 1: 55 % -1-1
01
1
NMS AXIS 3: 20%
G
A
B
A
B
G
A
B
G
B
GA
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
BG
BG
A
B
G
A
B
G
A
B
G
A
B
B
G
A
B
B
B
B
B
B
B
B
G A
B
B
G
A
B
B
B
G
A
A
A
G
B
PIMPLESMAIN DUNES MARSH
SHRUBSUMMIT
-1
0
1N
MS
AXI
S 2:
21%
-2-1
0NMS AXIS 1: 55 % -1-1
01
1
NMS AXIS 3: 20%
G
A
B
A
B
G
A
B
G
B
GA
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
BG
BG
A
B
G
A
B
G
A
B
G
A
B
B
G
A
B
B
B
B
B
B
B
B
G A
B
B
G
A
B
B
B
G
A
A
A
G
B
PIMPLESMAIN DUNES MARSH
SHRUBSUMMIT
FIG. 32. Non-metric multidimensional scaling ordination of pimple and dune transects based on environmental factors.
NM
S A
XIS
2: 2
1%
NMS AXIS 1: 55%NMS AXIS 3: 20%
-1
0
1
-2-1
0-1
0
1
G
A
B
A
B
G
A
B
G
B
GA
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
B
G
A
B
B
B
B
B
B
B
B
GA
B
B
G
A
B
B
B
G
A
A
A
G
B
PIMPLESMAIN DUNES MARSH
SHRUBSUMMIT
6 379 752
A
A
A
soil potassium (ppm)
NM
S A
XIS
2: 2
1%
NMS AXIS 1: 55%NMS AXIS 3: 20%
NM
S A
XIS
2: 2
1%
NMS AXIS 1: 55%NMS AXIS 3: 20%
-1
0
1
-2-1
0-1
0
1
G
A
B
A
B
G
A
B
G
B
GA
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
B
G
A
B
B
B
B
B
B
B
B
GA
B
B
G
A
B
B
B
G
A
A
A
G
B
PIMPLESMAIN DUNES MARSH
SHRUBSUMMIT
6 379 752
A
A
A
soil potassium (ppm)
FIG. 33. Overlay of potassium concentration on pimple and dune transects in a non-metric multidimensional scaling ordination.
87
-1
0
1N
MS
AXI
S 2:
21%
-2-1
0NMS AXIS 1: 55% -1
0
1
NMS AXIS 3: 20%
G
A
B
A
B
G
A
B
G
B
GA
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
BG
B
G
A
B
GA
B
G
A
B
G
A
B
B
G
A
B
B
B
B
B
B
B
B
G A
B
B
G
A
B
B
B
G
A
A
A
G
B
PIMPLESMAIN DUNES MARSH
SHRUBSUMMIT
-116 -45 25
A
A
A
water table position(cm)
-1
0
1N
MS
AXI
S 2:
21%
-2-1
0NMS AXIS 1: 55% -1
0
1
NMS AXIS 3: 20%
G
A
B
A
B
G
A
B
G
B
GA
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
BG
B
G
A
B
GA
B
G
A
B
G
A
B
B
G
A
B
B
B
B
B
B
B
B
G A
B
B
G
A
B
B
B
G
A
A
A
G
B
PIMPLESMAIN DUNES MARSH
SHRUBSUMMIT
-116 -45 25
A
A
A
water table position(cm)
FIG. 34. Overlay of water table height on pimple and main dune transects in a non-metric multidimensional scaling ordination.
-1
0
1
-2-1
0-1
0
1
G
A
B
A
B
G
A
B
G
B
GA
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
B
G
A
B
B
B
B
B
B
B
B
G A
B
B
G
A
B
B
B
G
A
A
A
G
B
PIMPLESMAIN DUNES MARSH
SHRUBSUMMIT
5 534 1062
A
A
A
soil ammonium (ppm)
NM
S AX
IS 2
: 21%
NMS AXIS 1: 55%NMS AXIS 3: 20%
-1
0
1
-2-1
0-1
0
1
G
A
B
A
B
G
A
B
G
B
GA
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
B
G
A
B
B
B
B
B
B
B
B
G A
B
B
G
A
B
B
B
G
A
A
A
G
B
PIMPLESMAIN DUNES MARSH
SHRUBSUMMIT
5 534 1062
A
A
A
soil ammonium (ppm)
NM
S AX
IS 2
: 21%
NMS AXIS 1: 55%NMS AXIS 3: 20%
NM
S AX
IS 2
: 21%
NMS AXIS 1: 55%NMS AXIS 3: 20%
Fig. 35. Overlay of soil ammonium concentration on pimple and dune transects in a non-metric multidimensional scaling ordination.
88
Transect ordinations: pimples with all recorded environmental factors
Canonical correspondence analysis of the transect species abundances versus
transect environmental variables for pimples described 31 % in the first three axes of the
solution (Table 13). This was an improvement over CCA ordination with both main dune
and pimple transects. Axis 1 represented an elevation – water table gradient (Fig. 37 &
Fig. 38). Axis 2 was correlated with magnesium (Fig. 39) and cation-exchange capacity
(Fig. 40). Axis 3 was correlated with phosphorus, magnesium saturation, organic matter,
and iron (Fig. 41–Fig. 42).
TABLE 13. Pearson’s r correlation coefficients for three axes of a CCA solution describing the relationship between pimple transect assemblages and environmental factors.
Axis 1: 12 % Axis 2: 10 % Axis 3: 9 % r r r elevation 0.906 Mg -0.434 P -0.795 water table -0.901 CEC -0.406 MgSat 0.694 o horizon -0.784 Ca -0.306 OM 0.646 CEC -0.54 salinity -0.304 Fe 0.621 B -0.534 OM -0.295 Zn -0.597 Mg -0.527 NH4 0.295 CaSat -0.589 Ca -0.482 B -0.27 Cu -0.463 Zn -0.425 water table -0.262 B -0.456 NOx 0.354 slope 0.253 Ca -0.412 Salinity -0.297 elevation 0.241 slope 0.372 Mn 0.246 Cu -0.238 CEC -0.223 P -0.237 CaSat 0.24 Mn -0.207 Cu -0.222 MgSat -0.217 NOx 0.141 Fe -0.215 Zn -0.208 salinity 0.139 K -0.189 Fe -0.176 water table -0.121 CaSat 0.166 K -0.162 elevation 0.12 slope 0.157 Mn 0.151 east 0.116 MgSat -0.154 P -0.145 o horizon 0.109 OM -0.147 north 0.128 KSat -0.063 East -0.081 KSat -0.058 Mg -0.023 North -0.084 o horizon -0.041 NH4 0.021 NH4 -0.077 east 0.038 K -0.016
89
-1
0
1
2
CC
A A
XIS
2: 1
0%
-1012
CCA AXIS 3: 9%
-1
0
1
2
CCA AX
IS 1:
12%
G
A
B
A
B
G
B
G
B
G
A
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
A MARSHG SHRUBB SUMMIT
elev
O horizon
east
north
salinity
slope
watertable
PK
CaMg
Zn
Mn
CuFe
B
CEC
CaSat
MgSat
KSat
OM
NH4
NOX
-1
0
1
2
CC
A A
XIS
2: 1
0%
-1012
CCA AXIS 3: 9%
-1
0
1
2
CCA AX
IS 1:
12%
G
A
B
A
B
G
B
G
B
G
A
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
A MARSHG SHRUBB SUMMIT
elev
O horizon
east
north
salinity
slope
watertable
PK
CaMg
Zn
Mn
CuFe
B
CEC
CaSat
MgSat
KSat
OM
NH4
NOX
FIG. 36. Canonical correspondence analysis of pimple transects with all recorded environmental variables.
-1
0
1
2
-1012-1
0
1
2
G
A
B
A
B
G
B
G
B
G
A
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
A MARSHG SHRUBB SUMMIT
0.0 0.8 1.5
A
A
A
elevation abovemarsh (m)
CC
A A
XIS
2: 1
0%
CCA AXIS 3: 9% CCA
AXIS
1: 12
%
-1
0
1
2
-1012-1
0
1
2
G
A
B
A
B
G
B
G
B
G
A
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
A MARSHG SHRUBB SUMMIT
0.0 0.8 1.5
A
A
A
elevation abovemarsh (m)
CC
A A
XIS
2: 1
0%
CCA AXIS 3: 9% CCA
AXIS
1: 12
%
CC
A A
XIS
2: 1
0%
CCA AXIS 3: 9% CCA
AXIS
1: 12
%
FIG. 37. Overlay of elevation above marsh on a CCA of pimple transects.
90
-1
-1
0
1
2
0 1 2-1
01
2
G
A B
A
B
G B
G
B
G
A
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
A MARSHG SHRUBB SUMMIT
0.00 58.00 116.00
A
A
A
water table(cm above min)
CC
A A
XIS
2: 1
0%
CCA AXIS 3: 9%
CCA AXIS 1: 12%
-1
-1
0
1
2
0 1 2-1
01
2
G
A B
A
B
G B
G
B
G
A
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
A MARSHG SHRUBB SUMMIT
0.00 58.00 116.00
A
A
A
water table(cm above min)
CC
A A
XIS
2: 1
0%
CCA AXIS 3: 9%
CCA AXIS 1: 12% FIG. 38. Overlay of water table position above the mean minimum level on a CCA of pimple transects.
-1
-1
0
1
2
0 1 2-1
01
2
G
AB
A
B
G B
G
B
G
A
G
A
B
GA
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
A MARSHG SHRUBB SUMMIT
24 149 274
A
A
A
soil magnesium (ppm)
CC
A A
XIS
2: 1
0%
CCA AXIS 3: 9%
CCA AXIS 1: 12%
-1
-1
0
1
2
0 1 2-1
01
2
G
AB
A
B
G B
G
B
G
A
G
A
B
GA
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
A MARSHG SHRUBB SUMMIT
24 149 274
A
A
A
soil magnesium (ppm)
CC
A A
XIS
2: 1
0%
CCA AXIS 3: 9%
CCA AXIS 1: 12% FIG. 39. Overlay of soil magnesium concentration on a CCA of pimple transects.
91
-1
-1
0
1
2
0 1 2-1
01
2
G
A
B
A
B
G B
G
B
G
A
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
A MARSHG SHRUBB SUMMIT
0.5 2.4 4.3
A
A
A
cation exchange capacity (meq / 100 g)
CC
A A
XIS
2: 1
0%
CCA AXIS 3: 9%
CCA AXIS 1: 12%
-1
-1
0
1
2
0 1 2-1
01
2
G
A
B
A
B
G B
G
B
G
A
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
A MARSHG SHRUBB SUMMIT
0.5 2.4 4.3
A
A
A
cation exchange capacity (meq / 100 g)
CC
A A
XIS
2: 1
0%
CCA AXIS 3: 9%
CCA AXIS 1: 12% FIG. 40. Overlay of cation-exchange capacity on a CCA of pimple transects.
-10
12 -1 0 1 2
-1
0
1
2
G
AB
A
B
G
B
G
B
G
A
G
A
B
G
A
B
G
A
BAB
G
A
B
G
A
B
G
G
A
B
G
B
G
AB
G
A
B
G
A
B
GA
B
A MARSHG SHRUBB SUMMIT
5 20 35
A
A
A
total soil P(ppm)
-10
12 -1 0 1 2
-1
0
1
2
G
AB
A
B
G
B
G
B
G
A
G
A
B
G
A
B
G
A
BAB
G
A
B
G
A
B
G
G
A
B
G
B
G
AB
G
A
B
G
A
B
GA
B
A MARSHG SHRUBB SUMMIT
5 20 35
A
A
A
total soil P(ppm)
FIG. 41. Overlay of soil phosphorus concentration on a CCA of pimple transects.
92
-10
12 -1 0 1 2
-1
0
1
2
GA
B
A
B
G
B
G
B
G
A
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
GA
B
G
A
B
G
A
B
A MARSHG SHRUBB SUMMIT
36 133 229
A
AA
soil iron (ppm)
CC
A A
XIS
2: 1
0%
CCA AXIS 3: 9%
CCA AXIS 1: 12%
-10
12 -1 0 1 2
-1
0
1
2
GA
B
A
B
G
B
G
B
G
A
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
GA
B
G
A
B
G
A
B
A MARSHG SHRUBB SUMMIT
36 133 229
A
AA
soil iron (ppm)
CC
A A
XIS
2: 1
0%
CCA AXIS 3: 9%
CCA AXIS 1: 12% FIG. 42. Overlay of soil iron concentration on a CCA of pimple transects.
A two-dimensional non-metric multidimensional scaling solution described
variation best, with the axes explaining 47 % and 48 % of variation in the data (Table
14). Shrub transects were not grouped with each other as distinctly as summit and marsh
transects (Fig. 43) Water table position, elevation, salinity, magnesium, and ammonium
were important to defining groups of transects (Fig. 46–Fig. 48).
Summary. For both ordination methods, important factors in explaining variation
between pimple transects were water table position, elevation above marsh, and thickness
of organic horizon. Boron, calcium, magnesium and phosphorus were more important in
the CCA than the NMS analysis. Salinity, ammonium, and potassium were more
important in the NMS solution than in CCA ordination. Transects were more distinctly
segregated by habitat in the CCA ordination than the NMS.
93
TABLE 14. Pearson’s r correlation coefficients for two axes in a non-metric multidimensional scaling ordination of pimple transects based on environmental factors.
Axis 1: 47 % Axis 2: 48 % r R elevation -0.567 salinity -0.856 water table 0.546 water table -0.549 NH4 -0.521 elevation 0.459 salinity 0.512 MgSat -0.427 K 0.479 CaSat 0.408 NOx -0.462 Fe -0.39 o horizon 0.447 Mn 0.366 CEC 0.443 o horizon -0.349 KSat 0.354 Mg -0.324 CaSat -0.351 B -0.233 Mg 0.34 CEC -0.219 Zn 0.324 Zn -0.214 Cu 0.281 OM -0.187 B 0.272 slope 0.132 Ca 0.243 Ca -0.093 slope -0.226 K -0.079 P 0.205 north 0.077 Mn -0.188 NH4 -0.06 MgSat 0.108 NOx 0.04 Fe 0.08 KSat -0.026 north -0.061 Cu -0.013 OM 0.048 P 0.009
A MARSHG SHRUBB SUMMIT
-2 -1 0 1
-1
0
1
G
A
B
A
B
G
A
B
G
B
G
A
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
NMS AXIS 1: 47%
NM
S A
XIS
2: 4
8%
A MARSHG SHRUBB SUMMIT
-2 -1 0 1
-1
0
1
G
A
B
A
B
G
A
B
G
B
G
A
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
NMS AXIS 1: 47%
NM
S A
XIS
2: 4
8%
FIG. 43. Non-metric multidimensional scaling ordination of pimple transects.
94
NMS AXIS 1: 47%
NM
S A
XIS
2: 4
8%
A MARSHG SHRUBB SUMMIT
0 58 116
A
A
A
water table heightabove min (cm)
-2 -1 0 1
-1
0
1
G
A
B
A
B
G
A
B
G
B
G
A
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
NMS AXIS 1: 47%
NM
S A
XIS
2: 4
8%
A MARSHG SHRUBB SUMMIT
0 58 116
A
A
A
water table heightabove min (cm)
-2 -1 0 1
-1
0
1
G
A
B
A
B
G
A
B
G
B
G
A
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
FIG. 44 Overlay of water table position on a NMS ordination of pimple transects.
NMS AXIS 1: 47%
NM
S A
XIS
2: 4
8%
A MARSHG SHRUBB SUMMIT
0 .75 1.5
A
A
A
elevation above marsh (m)
-2 -1 0 1
-1
0
1
G
A
B
A
B
G
A
B
G
B
G
A
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
BG
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
NMS AXIS 1: 47%
NM
S A
XIS
2: 4
8%
A MARSHG SHRUBB SUMMIT
0 .75 1.5
A
A
A
elevation above marsh (m)
-2 -1 0 1
-1
0
1
G
A
B
A
B
G
A
B
G
B
G
A
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
BG
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
FIG. 45. Graphic overlay of elevation on pimple transects in an NMS ordination.
95
A MARSHG SHRUBB SUMMIT
7 1594 3180
A
A
Asalinity (ppm)
-2 -1 0 1
-1
0
1
G
A
B
A
B
G
A
B
G
B
G
A
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
NMS AXIS 1: 47%
NM
S A
XIS
2: 4
8%
A MARSHG SHRUBB SUMMIT
7 1594 3180
A
A
Asalinity (ppm)
-2 -1 0 1
-1
0
1
G
A
B
A
B
G
A
B
G
B
G
A
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
NMS AXIS 1: 47%
NM
S A
XIS
2: 4
8%
FIG. 46. Overlay of salinity on pimple transects in an NMS ordination.
NMS AXIS 1: 47%
NM
S A
XIS
2: 4
8%
A MARSHG SHRUBB SUMMIT
6 534 1062
A
AA
soil ammonium(ppm)
-2 -1 0 1
-1
0
1
G
A
B
A
B
G
A
B
G
B
G
A
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
NMS AXIS 1: 47%
NM
S A
XIS
2: 4
8%
A MARSHG SHRUBB SUMMIT
6 534 1062
A
AA
soil ammonium(ppm)
-2 -1 0 1
-1
0
1
G
A
B
A
B
G
A
B
G
B
G
A
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
FIG. 47. Overlay of ammonium concentrations on pimple transects in an NMS ordination.
96
NMS AXIS 1: 47%
NM
S AX
IS 2
: 48%
A MARSHG SHRUBB SUMMIT
24 149 274
A
AA
soil magnesium (ppm)
-2 -1 0 1
-1
0
1
G
A
B
A
B
G
A
B
G
B
G
A
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
NMS AXIS 1: 47%
NM
S AX
IS 2
: 48%
A MARSHG SHRUBB SUMMIT
24 149 274
A
AA
soil magnesium (ppm)
-2 -1 0 1
-1
0
1
G
A
B
A
B
G
A
B
G
B
G
A
G
A
B
G
A
B
G
A
B
A
B
G
A
B
G
A
B
G
G
A
B
G
B
G
A
B
G
A
B
G
A
B
G
A
B
FIG. 48. Overlay of magnesium concentration on pimple transects in an NMS ordination.
Species ordinations: species from both main dunes and pimples
Overall, ordinations of species based on environmental variables (the mean from
transects in which the species occurred) explained more variation than ordinations of
transects based on environmental variables. Surprisingly, mean water table height did not
have strong correlation coefficients on most major axes of most ordinations of species.
Canonical correspondence analysis of data combined from both dune and pimple
species explained 26 % of variation in the first three axes (Table 15; Fig. 49). The three
axes were highly correlated with boron, zinc, and magnesium; iron; and organic matter
and ammonium, respectively (Fig. 50 – Fig. 55). Based on previous observation and an
initial examination of the results, I tracked a few species through overlays of
environmental factors on the dune and pimple CCA.
97
Five species represented marsh and summit flora. Ammophila breviligulata and
Panicum amarum are both dominant grasses of dry dune summits. Cyperus strigosus and
Distichlis spicata are representative hydrophytic graminoids. Spartina patens was
common in both xeric and hydric environments. D. spicata is apparently at a competitive
advantage over S. patens in hypersaline conditions, but is otherwise competitively
inferior (Bertness and Ellison 1987, Bertness 1991a, Costa et al. 2003). The xerophytic
and hydrophytic pairs differed most in their association with nutrients such as boron (Fig.
50) and magnesium (Fig. 52) and in affinity for fresh water (Fig. 56). Spartina patens,
which has been shown to have a low tolerance for chronic inundation, was nevertheless
either similar in environmental factor affinity to hydrophytes or intermediate (Bertness
1991b).
Both Hydrocotyle verticellata and Centella erecta are members of the carrot
family, Apiaceae; stoloniferous, low-growing herbs; similar in appearance; and reputed to
be hydrophytic (Radford et al. 1968). Nevertheless, I observed in the field that they did
not occur near each other. Regardless of being hydrophytes, both species were different
with regard to most environmental factors (Fig. 50, e.g.; Fig. 56).
Aralia spinosa is a small (< 4 m) tree in the ivy family, Araliaceae, which is
considered allied to or part of the Apiaceae. I only found it growing in the same area as
Centella erecta. For most factors, it was either similar in affinity to C. erecta or
intermediate between C. erecta and Hydrocotlye verticellata. A notable exception was
the high soil ammonium found with A. spinosa (Fig. 55).
I compared the grass Festuca rubra and the tree Juniperus virginiana based on
my a priori observations that 1) they both seemed to have an affinity for each other and
98
2) F. rubra was nearly always found in the margin of the shrub – summit interface. Both
species were similar in affinity to most factors.
TABLE 15. Pearson’s r correlation coefficients between environmental factors and three axes of a canonical correspondence analysis ordination of dune and pimple species.
Axis 1: 11 % Axis 2: 9 % Axis 3: 6 % r r r B -0.871 Fe -0.736 OM -0.305 Zn -0.842 MgSat -0.542 NH4 0.2 Mg -0.82 OM -0.523 Mg -0.19 Fe -0.737 CaSat 0.279 Ca -0.177 CEC -0.701 Mg -0.275 CEC -0.174 Ca -0.634 Mn 0.276 Mn -0.155 P -0.633 CEC -0.203 Zn -0.136 OM -0.575 Ca -0.185 P 0.118 MgSat -0.55 NOx 0.18 NOx -0.101 Cu -0.487 KSat 0.165 water table 0.075 NOx 0.417 P 0.159 Fe -0.059 Mn 0.387 water table 0.125 K 0.056 water table -0.361 Cu 0.119 B -0.04 KSat 0.294 Zn -0.102 CaSat -0.03 CaSat 0.119 K 0.095 Cu 0.025 K 0.094 B -0.089 MgSat 0.011 NH4 -0.022 NH4 -0.06 KSat 0.016
A HYDRICG MESICB XERIC
-2 -1 0 1 2 3CCA AXIS 1: 11%
-1
0
1
2
CC
A A
XIS
2: 9
%
B
B
B
B
A
A
G
A
B
B
A
A
G
B
G
B
G
A
A
A
G
G
A A
G
G
A
B
B
G
B
G
B
B
G
G
G
A
A
AA
G
A
G
G
A
B
A
B
G
A
G
A
A
water
P
KCa
Mg
Zn MnCu
Fe
BNH4 NO3
A HYDRICG MESICB XERIC
-2 -1 0 1 2 3CCA AXIS 1: 11%
-1
0
1
2
CC
A A
XIS
2: 9
%
B
B
B
B
A
A
G
A
B
B
A
A
G
B
G
B
G
A
A
A
G
G
A A
G
G
A
B
B
G
B
G
B
B
G
G
G
A
A
AA
G
A
G
G
A
B
A
B
G
A
G
A
A
water
P
KCa
Mg
Zn MnCu
Fe
BNH4 NO3
FIG. 49. CCA of main dune and pimple species based on mean environmental variables. In this and subsequent figures, the least important axis was omitted for clarity.
99
A HYDRICG MESICB XERIC
0.1 0.5 1.0
A
AA
soil boron (ppm)
-2 -1 0 1 2 3CCA AXIS 1: 11%
-1
0
1
2
CC
A A
XIS
2: 9
%B
B
B
B
A
A
G
A
B
B
A
A
G
B
G
B
G
A
A
A
G
G
AA
G
G
A
B
B
G
B
G
B
B
G
G
G
A
A
AA
G
A
G
G
A
B
A
B
G
A
G
A
A
Centella erecta
Cyperus strigosus
Hydrocotyle verticellata
Spartina patens
Festuca rubraJuniperus virginiana
Distichlis spicataPanicum amarum
Ammophila brevilgulata
Aralia spinosa
A HYDRICG MESICB XERIC
0.1 0.5 1.0
A
AA
soil boron (ppm)
-2 -1 0 1 2 3CCA AXIS 1: 11%
-1
0
1
2
CC
A A
XIS
2: 9
%B
B
B
B
A
A
G
A
B
B
A
A
G
B
G
B
G
A
A
A
G
G
AA
G
G
A
B
B
G
B
G
B
B
G
G
G
A
A
AA
G
A
G
G
A
B
A
B
G
A
G
A
A
Centella erecta
Cyperus strigosus
Hydrocotyle verticellata
Spartina patens
Festuca rubraJuniperus virginiana
Distichlis spicataPanicum amarum
Ammophila brevilgulata
Aralia spinosa
Centella erecta
Cyperus strigosus
Hydrocotyle verticellata
Spartina patens
Festuca rubraJuniperus virginiana
Distichlis spicataPanicum amarum
Ammophila brevilgulata
Aralia spinosa
FIG. 50. Overlay of soil boron on a CCA ordination of main dune and pimple species.
A HYDRICG MESICB XERIC
0 1 2
AA
A
soil zinc (ppm)
-2 -1 0 1 2 3CCA AXIS 1: 11%
-1
0
1
2
CC
A A
XIS
2: 9
%
B
B
B
B
A
A
G
A
B
B
A
A
G
B
G
B
G
A
A
A
G
G
AA
G
G
A
B
B
G
B
G
B
B
G
G
G
A
A
AA
G
A
G
G
A
B
A
B
G
A
G
A
A
Centella erecta
Cyperus strigosus
Hydrocotyle verticellata
Spartina patens
Festuca rubraJuniperus virginiana
Distichlis spicataPanicum amarum
Ammophila brevilgulata
Aralia spinosa
A HYDRICG MESICB XERIC
0 1 2
AA
A
soil zinc (ppm)
-2 -1 0 1 2 3CCA AXIS 1: 11%
-1
0
1
2
CC
A A
XIS
2: 9
%
B
B
B
B
A
A
G
A
B
B
A
A
G
B
G
B
G
A
A
A
G
G
AA
G
G
A
B
B
G
B
G
B
B
G
G
G
A
A
AA
G
A
G
G
A
B
A
B
G
A
G
A
A
Centella erecta
Cyperus strigosus
Hydrocotyle verticellata
Spartina patens
Festuca rubraJuniperus virginiana
Distichlis spicataPanicum amarum
Ammophila brevilgulata
Aralia spinosa
Centella erecta
Cyperus strigosus
Hydrocotyle verticellata
Spartina patens
Festuca rubraJuniperus virginiana
Distichlis spicataPanicum amarum
Ammophila brevilgulata
Aralia spinosa
FIG. 51. Overlay of soil zinc on a CCA ordination of dune and pimple species.
100
A HYDRICG MESICB XERIC
-1 124 249
A
A
A
soil Mg (ppm)
-2 -1 0 1 2 3CCA AXIS 1: 11%
-1
0
1
2
CC
A A
XIS
2: 9
%B
B
B
B
A
A
G
A
B
B
A
A
G
B
G
B
G
A
A
A
G
G
AA
G
G
A
B
B
G
B
G
B
B
G
G
G
A
A
AA
G
A
G
G
A
B
A
B
G
A
G
A
A
Centella erecta
Cyperus strigosus
Hydrocotyle verticellata
Spartina patens
Festuca rubraJuniperus virginiana
Distichlis spicataPanicum amarum
Ammophila brevilgulata
Aralia spinosa
A HYDRICG MESICB XERIC
-1 124 249
A
A
A
soil Mg (ppm)
-2 -1 0 1 2 3CCA AXIS 1: 11%
-1
0
1
2
CC
A A
XIS
2: 9
%B
B
B
B
A
A
G
A
B
B
A
A
G
B
G
B
G
A
A
A
G
G
AA
G
G
A
B
B
G
B
G
B
B
G
G
G
A
A
AA
G
A
G
G
A
B
A
B
G
A
G
A
A
Centella erecta
Cyperus strigosus
Hydrocotyle verticellata
Spartina patens
Festuca rubraJuniperus virginiana
Distichlis spicataPanicum amarum
Ammophila brevilgulata
Aralia spinosa
Centella erecta
Cyperus strigosus
Hydrocotyle verticellata
Spartina patens
Festuca rubraJuniperus virginiana
Distichlis spicataPanicum amarum
Ammophila brevilgulata
Aralia spinosa
FIG. 52. Overlay of soil magnesium on a CCA ordination of dune and pimple species.
A HYDRICG MESICB XERIC
0 67 133
AAA
soil iron (ppm)
-2 -1 0 1 2 3CCA AXIS 1: 11%
-1
0
1
2
CC
A A
XIS
2: 9
%
B
B
B
B
A
A
G
A
B
B
A
A
G
B
G
B
G
A
A
A
G
G
AA
G
G
A
B
B
G
B
G
B
B
G
G
G
A
A
AA
G
A
G
G
A
B
A
B
G
A
G
A
A
Centella erecta
Cyperus strigosus
Hydrocotyle verticellata
Spartina patens
Festuca rubraJuniperus virginiana
Distichlis spicata Panicum amarum
Ammophila brevilgulata
Aralia spinosa
A HYDRICG MESICB XERIC
0 67 133
AAA
soil iron (ppm)
-2 -1 0 1 2 3CCA AXIS 1: 11%
-1
0
1
2
CC
A A
XIS
2: 9
%
B
B
B
B
A
A
G
A
B
B
A
A
G
B
G
B
G
A
A
A
G
G
AA
G
G
A
B
B
G
B
G
B
B
G
G
G
A
A
AA
G
A
G
G
A
B
A
B
G
A
G
A
A
Centella erecta
Cyperus strigosus
Hydrocotyle verticellata
Spartina patens
Festuca rubraJuniperus virginiana
Distichlis spicata Panicum amarum
Ammophila brevilgulata
Aralia spinosa
FIG. 53. Overlay of soil iron on a CCA ordination of dune and pimple species.
101
A HYDRICG MESICB XERIC
-1 2 4
A
A
A
organic matter (%)
-2 -1 0 1 2 3CCA AXIS 1: 11%
-1
0
1
2
CC
A A
XIS
2: 9
%B
B
B
B
AA
G
A
B
B
A
A
G
B
G
B
G
A
A
A
G
G
AA
G
G
A
B
B
G
B
G
B
B
G
G
G
A
A
A
A
G
A
G
G
A
B
A
B
G
A
G
A
A
Centella erecta
Cyperus strigosus
Hydrocotyle verticellata
Spartina patens
Festuca rubraJuniperus virginiana
Distichlis spicataPanicum amarum
Ammophila brevilgulata
Aralia spinosa
A HYDRICG MESICB XERIC
-1 2 4
A
A
A
organic matter (%)
-2 -1 0 1 2 3CCA AXIS 1: 11%
-1
0
1
2
CC
A A
XIS
2: 9
%B
B
B
B
AA
G
A
B
B
A
A
G
B
G
B
G
A
A
A
G
G
AA
G
G
A
B
B
G
B
G
B
B
G
G
G
A
A
A
A
G
A
G
G
A
B
A
B
G
A
G
A
A
Centella erecta
Cyperus strigosus
Hydrocotyle verticellata
Spartina patens
Festuca rubraJuniperus virginiana
Distichlis spicataPanicum amarum
Ammophila brevilgulata
Aralia spinosa
FIG. 54. Overlay of soil organic matter on a CCA ordination of dune and pimple species.
A HYDRICG MESICB XERIC
0 206 412
A
A
A
soil ammonium (ppm)
-2 -1 0 1 2 3CCA AXIS 1: 11%
-1
0
1
2
CC
A A
XIS
2: 9
% B
B
B
B
A
A
G
A
B
B
A
A
G
B
G
B
G
A
A
A
G
G
A A
G
G
A
B
B
G
B
G
B
B
G
G
G
A
A
A
A
G
A
G
G
A
B
A
B
G
A
G
A
A
Centella erecta
Cyperus strigosus
Hydrocotyle verticellata
Spartina patens
Festuca rubraJuniperus virginiana
Distichlis spicataPanicum amarum
Ammophila brevilgulata
Aralia spinosa
A HYDRICG MESICB XERIC
0 206 412
A
A
A
soil ammonium (ppm)
-2 -1 0 1 2 3CCA AXIS 1: 11%
-1
0
1
2
CC
A A
XIS
2: 9
% B
B
B
B
A
A
G
A
B
B
A
A
G
B
G
B
G
A
A
A
G
G
A A
G
G
A
B
B
G
B
G
B
B
G
G
G
A
A
A
A
G
A
G
G
A
B
A
B
G
A
G
A
A
Centella erecta
Cyperus strigosus
Hydrocotyle verticellata
Spartina patens
Festuca rubraJuniperus virginiana
Distichlis spicataPanicum amarum
Ammophila brevilgulata
Aralia spinosa
Centella erecta
Cyperus strigosus
Hydrocotyle verticellata
Spartina patens
Festuca rubraJuniperus virginiana
Distichlis spicataPanicum amarum
Ammophila brevilgulata
Aralia spinosa
FIG. 55. Overlay of soil ammonium on a CCA ordination of dune and pimple species.
102
A HYDRICG MESICB XERIC
0 12 24
A
A A
water table height above min. (cm)
-2 -1 0 1 2 3CCA AXIS 1: 11%
-1
0
1
2
CC
A A
XIS
2: 9
% B
B
B
B
A
A
G
A
B
B
A
A
G
B
G
B
G
A
A
A
G
G
AA
G
G
A
B
B
G
B
G
B
B
G
G
G
A
A
AA
G
A
G
G
A
B
A
B
G
A
G
A
A
Centella erecta
Cyperus strigosus
Hydrocotyle verticellata
Spartina patens
Festuca rubraJuniperus virginiana
Distichlis spicataPanicum amarum
Ammophila brevilgulata
Aralia spinosa
A HYDRICG MESICB XERIC
0 12 24
A
A
0 12 24
A
A A
water table height above min. (cm)
-2 -1 0 1 2 3CCA AXIS 1: 11%
-1
0
1
2
CC
A A
XIS
2: 9
% B
B
B
B
A
A
G
A
B
B
A
A
G
B
G
B
G
A
A
A
G
G
AA
G
G
A
B
B
G
B
G
B
B
G
G
G
A
A
AA
G
A
G
G
A
B
A
B
G
A
G
A
A
Centella erecta
Cyperus strigosus
Hydrocotyle verticellata
Spartina patens
Festuca rubraJuniperus virginiana
Distichlis spicataPanicum amarum
Ammophila brevilgulata
Aralia spinosa
Centella erecta
Cyperus strigosus
Hydrocotyle verticellata
Spartina patens
Festuca rubraJuniperus virginiana
Distichlis spicataPanicum amarum
Ammophila brevilgulata
Aralia spinosa
FIG. 56. Overlay of mean position of water table above the average minimum on a CCA ordination of dune and pimple species.
A three-dimensional solution to a non-metric multidimensional scaling ordination
of dune and pimple species explained 98 % of variation in the species – environmental
factor matrix (Table 16). In contrast to the CCA, species did not form obvious groups
based on my a priori fresh water-affinity designations (Fig. 57). The two most-correlated
factors to each axis were NOx concentration and potassium base saturation (Fig. 57 &
Fig. 58); calcium base saturation and iron; and cation-exchange capacity and magnesium
(Fig. 59).
103
TABLE 16. Pearson’s r correlation coefficients between environmental factors and three axes of a non-metric multidimensional scaling ordination of pimple and dune species.
Axis 1: 66 % Axis 2: 8 % Axis 3: 24 % r r r NOx -0.826 CaSat 0.365 CEC 0.913 KSat -0.798 Fe 0.291 Mg 0.878 K -0.696 NH4 0.248 B 0.845 CaSat 0.572 K 0.22 Ca 0.817 Mg 0.565 Mn 0.212 OM 0.761 B 0.557 KSat 0.106 Zn 0.756 Ca 0.547 Zn 0.093 Fe 0.46 MgSat 0.51 P 0.074 P 0.405 Zn 0.481 Cu 0.044 NH4 0.252 NH4 -0.438 B 0.04 K 0.25 CEC 0.42 MgSat 0.039 water table 0.228 Cu 0.395 CEC 0.035 Cu 0.218 OM 0.395 Mg 0.029 CaSat -0.168 Fe 0.389 OM 0.02 MgSat 0.088 P 0.351 water table 0.017 KSat 0.068 water table 0.318 NOx 0.004 Mn -0.014 Mn -0.248 Ca 0.001 NOx -0.017
A HYDRICG MESICB XERIC
1 122 242
A
A A
NOx (ppm)
-2 -1 0 1
NMS AXIS 1: 66 %
-1
0
1
2
NM
S A
XIS
3: 2
4 %
B
B
B
B
AA
G
A
B
B
A
A
G
B
G
B
G
A
A
A
G
G
A
A
G
G
A
B
B
G
B
G
B
B
G
G
G
AA
A A
G
A
G
G
A
B
A
B
G
A
G
A
A
A HYDRICG MESICB XERIC
1 122 242
A
A A
NOx (ppm)
-2 -1 0 1
NMS AXIS 1: 66 %
-1
0
1
2
NM
S A
XIS
3: 2
4 %
B
B
B
B
AA
G
A
B
B
A
A
G
B
G
B
G
A
A
A
G
G
A
A
G
G
A
B
B
G
B
G
B
B
G
G
G
AA
A A
G
A
G
G
A
B
A
B
G
A
G
A
A
FIG. 57. Overlay of NOx concentration on pimple and dune species in NMS ordination.
104
A HYDRICG MESICB XERIC
0 38 77
A
A
A
K saturation (%)
-2 -1 0 1
NMS AXIS 1: 66 %
-1
0
1
2
NM
S A
XIS
3: 2
4 %
B
B
B
B
A
A
G
A
B
B
A
A
G
B
G
B
G
A
A
A
G
G
A
A
G
G
A
B
B
G
B
G
B
B
G
G
G
A A
A A
G
A
G
G
A
B
A
B
G
A
G
A
A
A HYDRICG MESICB XERIC
0 38 77
A
A
A
K saturation (%)
-2 -1 0 1
NMS AXIS 1: 66 %
-1
0
1
2
NM
S A
XIS
3: 2
4 %
B
B
B
B
A
A
G
A
B
B
A
A
G
B
G
B
G
A
A
A
G
G
A
A
G
G
A
B
B
G
B
G
B
B
G
G
G
A A
A A
G
A
G
G
A
B
A
B
G
A
G
A
A
FIG. 58. Overlay of potassium base saturation on pimple and dune species in an NMS ordination.
A HYDRICG MESICB XERIC
-1 124 249
A
A
A
soil Mg (ppm)
NMS AXIS 1: 66 %
NM
S A
XIS
3: 2
4 %
B
B
B
B
AA
G
A
B
A
A
G
G
B
G
A
A
A
G
G
A
A
G
G
A
B
B
G
B
G
B
B
G
G
G
AA
A A
G
A
G
G
A
B
A
B
G
A
G
A
A
A HYDRICG MESICB XERIC
-1 124 249
A
A
A
soil Mg (ppm)
NMS AXIS 1: 66 %
NM
S A
XIS
3: 2
4 %
B
B
B
B
AA
G
A
B
A
A
G
G
B
G
A
A
A
G
G
A
A
G
G
A
B
B
G
B
G
B
B
G
G
G
AA
A A
G
A
G
G
A
B
A
B
G
A
G
A
A
FIG. 59. Overlay of magnesium concentration on dune and pimple species in an NMS ordination.
105
Summary. As with the ordinations of pimple and dune transects, boron,
magnesium, calcium, and cation-exchange capacity explained the most variation among
dune and pimple species. Also similar to the transect ordinations is the difference in the
importance of potassium between NMS and CCA. Water table was less important to
species ordinations than to transect ordinations, and organic matter content was of greater
importance in species ordinations.
Species ordinations: pimple species and all recorded factors
Canonical correspondence analysis of pimple species abundances in transects
versus mean environmental variables for each species explained 32 % of variation with
the first three axes (Fig. 60; Table 17). Some of the same variables that were important in
explaining species distributions in the combined pimple – main dune data were still
important, e.g., potassium, ammonium, and iron (Fig. 61 – Fig. 63). Water table position
was more important in pimple only CCA, and boron was much less important.
A two-dimensional non-metric multidimensional scaling ordination of pimple
species explained 93 % of variation in the data (Table 18; Fig. 64). Potassium
concentration and base saturation were most correlated to the first axis (Fig. 65), and
ammonium concentration and cation-exchange capacity were most important to the
second (Fig. 66). Fresh water availability was of intermediate importance.
Summary. Most of the same factors that were important to describing variation
among pimple transects were also important to explaining variation among pimple
species, with some exceptions. Potassium, of intermediate importance to explaining
pimple transects, was the most important factor for describing variation among pimple
species. Although among the factors describing the most variation, water table and
106
elevation were not as important in ordinations of species as they were in ordinations of
transects.
I weighted the Pearson’s r correlation coefficients of the factors by the
explanatory value of the axes in all the ordinations. Potassium, cation-exchange capacity,
water table position, calcium, magnesium and boron were most important in explaining
variation in the data. Phosphorus, nitrogen (especially as NH4), manganese, copper,
slope, and aspect were the least important.
TABLE 17. Pearson’s r correlation coefficients between environmental factors and three axes of a canonical correspondence analysis ordination of pimple species.
Axis 1: 12 % Axis 2: 11 % Axis 3: 9 % r r r K 0.567 NH4 0.536 Fe -0.38 KSat 0.552 O horizon 0.522 OM -0.345 MgSat -0.536 water table 0.451 Mn -0.305 O horizon 0.486 Elevation -0.447 slope 0.268 CEC 0.432 Mn -0.379 NOx 0.237 Fe -0.387 Slope 0.326 Zn -0.214 Slope 0.381 Zn 0.246 elevation 0.212 NH4 0.368 MgSat 0.242 CaSat -0.183 Zn -0.366 K -0.229 Ca -0.18 OM -0.348 KSat -0.225 B -0.176 Ca 0.256 Fe 0.211 salinity -0.17 Mn 0.252 B 0.182 O horizon -0.158 Cu -0.211 East -0.167 water table -0.148 East 0.184 North 0.136 K 0.125 NOx -0.181 Mg 0.087 KSat 0.123 Mg 0.18 NOx -0.08 north -0.105 Salinity -0.095 Ca 0.074 Mg -0.088 water table 0.088 Cu 0.058 NH4 0.076 Elevation -0.082 CEC -0.039 MgSat 0.064 B 0.082 Salinity 0.048 east -0.061 North -0.047 P 0.05 CEC -0.051 P 0.031 OM 0.036 Cu -0.045 CaSat -0.01 CaSat -0.017 P -0.021
107
-20
2
CCA AXIS 1: 12%-4
-20
2
CCA AXIS 3: 9%
-4
-2
0
2
4
CC
A A
XIS
2: 1
1%
B
G
A
A
G
BA
A
G
B
B
G
A
A
A
G
GA
A
G
G
A
A
B
G
A
B
G
GG
GB
G
G
A
A
A
A
G
A
GG
A
B
A
G
G
A
G
A
A
G
A HYDRICG MESICB XERIC
elevation
O horizon
east
north
salinity
slope
water table
P
K
Ca
Mg
Zn
KSatOM
NH4
NOx
-20
2
CCA AXIS 1: 12%-4
-20
2
CCA AXIS 3: 9%
-4
-2
0
2
4
CC
A A
XIS
2: 1
1%
B
G
A
A
G
BA
A
G
B
B
G
A
A
A
G
GA
A
G
G
A
A
B
G
A
B
G
GG
GB
G
G
A
A
A
A
G
A
GG
A
B
A
G
G
A
G
A
A
G
A HYDRICG MESICB XERIC
elevation
O horizon
east
north
salinity
slope
water table
P
K
Ca
Mg
Zn
KSatOM
NH4
NOx
elevation
O horizon
east
north
salinity
slope
water table
P
K
Ca
Mg
Zn
KSatOM
NH4
NOx
FIG. 60. Canonical correspondence analysis of pimple species based on environmental factors.
108
-4
-2
0
2
4
-20
24 -4
-20
2
B
G
A
A
G
B
A
A
G
B
B
G
A
A
A
G
G A
A
G
G
A
A
B
G
A
B
G
GG
G B
G
G
A
A
A
A
G
A
GG
A
B
A
G
G
A
G
A
A
G
A HYDRICG MESICB XERIC
9 273 538
A
A
A
soil potassium(ppm)
CC
A A
XIS
2: 1
1%
CCA AXIS 1: 12%CCA AXIS 3: 9%
-4
-2
0
2
4
-20
24 -4
-20
2
B
G
A
A
G
B
A
A
G
B
B
G
A
A
A
G
G A
A
G
G
A
A
B
G
A
B
G
GG
G B
G
G
A
A
A
A
G
A
GG
A
B
A
G
G
A
G
A
A
G
A HYDRICG MESICB XERIC
9 273 538
A
A
A
soil potassium(ppm)
CC
A A
XIS
2: 1
1%
CCA AXIS 1: 12%CCA AXIS 3: 9%
FIG. 61. Overlay of mean soil potassium on pimple species in a CCA ordination.
109
-4
-2
0
2
4
-20
24 -4
-20
2
B
G
A
A
G
B
A
A
G
B
B
G
A
A
A
G
G
A
A
G
G
A
A
B
G
A
B
G
GG
G B
G
G
A
A
A
A
G
A
GG
A
B
A
G
G
A
G
A
A
G
A HYDRICG MESICB XERIC
6 216 427
A
A
A
soil ammonium(ppm)
CC
A A
XIS
2: 1
1%
CCA AXIS 1: 12%CCA AXIS 3: 9%
-4
-2
0
2
4
-20
24 -4
-20
2
B
G
A
A
G
B
A
A
G
B
B
G
A
A
A
G
G
A
A
G
G
A
A
B
G
A
B
G
GG
G B
G
G
A
A
A
A
G
A
GG
A
B
A
G
G
A
G
A
A
G
A HYDRICG MESICB XERIC
6 216 427
A
A
A
soil ammonium(ppm)
CC
A A
XIS
2: 1
1%
CCA AXIS 1: 12%CCA AXIS 3: 9%
FIG. 62. Overlay of soil ammonium on pimple species in a CCA ordination.
110
-4
-2
0
2
4
-20
24 -4
-20
2
B
G
A
A
G
BA
A
G
B
B
G
A
A
A
G
G
A
A
G
G
A
A
B
G
A
B
G
G G
GB
G
G
A
A
A
A
G
A
GG
A
B
A
G
G
A
G
A
A
G
A HYDRICG MESICB XERIC
40 126 213
A
A
A
soil iron (ppm)
CC
A A
XIS
2: 1
1%
CCA AXIS 1: 12%CCA AXIS 3: 9%
-4
-2
0
2
4
-20
24 -4
-20
2
B
G
A
A
G
BA
A
G
B
B
G
A
A
A
G
G
A
A
G
G
A
A
B
G
A
B
G
G G
GB
G
G
A
A
A
A
G
A
GG
A
B
A
G
G
A
G
A
A
G
A HYDRICG MESICB XERIC
40 126 213
A
A
A
soil iron (ppm)
CC
A A
XIS
2: 1
1%
CCA AXIS 1: 12%CCA AXIS 3: 9%
FIG. 63. Overlay of mean soil iron concentrations on pimple species in a CCA ordination.
111
TABLE 18. Pearson’s r correlation coefficients between environmental factors and three axes of a non-metric multidimensional scaling ordination of pimple species.
Axis 1: 65 % Axis 2: 28 % r r K -0.846 NH4 -0.673 KSat -0.81 CEC 0.669 CEC -0.661 Ca 0.587 MgSat 0.655 Mg 0.574 O horizon -0.577 B 0.555 Fe 0.457 P 0.513 Ca -0.419 Zn 0.509 B -0.406 Cu 0.462 water table -0.38 K 0.459 elevation 0.378 MgSat -0.429 OM 0.359 NOx -0.428 Mg -0.345 KSat 0.422 NH4 -0.319 slope -0.387 slope -0.267 water table 0.33 P -0.24 elevation -0.303 NOx 0.184 east 0.198 Mn -0.168 north -0.115 Cu -0.112 O horizon -0.098 CaSat 0.078 Mn 0.078 east -0.03 OM -0.064 north -0.03 CaSat 0.055 Zn 0.035 Fe -0.041
A HYDRICG MESICB XERIC
0
NMS AXIS 1: 65 %
0
NM
S A
XIS
2: 2
8 %
B
G
A
A
G
B
A
A
G
B
B
G
A
A
A
G
G
A
A
G
G
A
B
G
A
B
G
G
G
G
B
G
G
A
A
AA
G
A
G
G
A
B
A
G
G
G
A
A
A HYDRICG MESICB XERIC
0
NMS AXIS 1: 65 %
0
NM
S A
XIS
2: 2
8 %
B
G
A
A
G
B
A
A
G
B
B
G
A
A
A
G
G
A
A
G
G
A
B
G
A
B
G
G
G
G
B
G
G
A
A
AA
G
A
G
G
A
B
A
G
G
G
A
A
FIG. 64. Two-axis non-metric multidimensional scaling ordination of pimple species based on environmental variables.
112
A HYDRICG MESICB XERIC
9 273 538
A
A
A
soil K (ppm)
-2 -1 0 1NMS AXIS 1: 65 %
-2
-1
0
1
NM
S A
XIS
2: 2
8 %
B
G
A
A
G
B
A
A
G
B
B
G
A
A
A
G
G
A
A
G
G
A
B
G
A
B
G
G
G
G
B
G
G
A
A
AA
G
A
G
G
A
B
A
G
G
G
A
A
A HYDRICG MESICB XERIC
9 273 538
A
A
A
soil K (ppm)
-2 -1 0 1NMS AXIS 1: 65 %
-2
-1
0
1
NM
S A
XIS
2: 2
8 %
B
G
A
A
G
B
A
A
G
B
B
G
A
A
A
G
G
A
A
G
G
A
B
G
A
B
G
G
G
G
B
G
G
A
A
AA
G
A
G
G
A
B
A
G
G
G
A
A
FIG. 65. Overlay of potassium concentration on pimples in an NMS ordination.
A HYDRICG MESICB XERIC
0.70 1.97 3.25
A
A
A
cation-exchangecapacity(meq/100 g)
-2 -1 0 1NMS AXIS 1: 65 %
-2
-1
0
1
NM
S A
XIS
2: 2
8 %
B
G
A
A
G
B
A
A
G
B
B
G
A
A
A
G
G
A
A
G
G
A
B
G
A
B
G
G
G
G
B
G
G
A
A
AA
G
A
G
G
A
B
A
G
G
G
A
A
A HYDRICG MESICB XERIC
0.70 1.97 3.25
A
A
A
cation-exchangecapacity(meq/100 g)
-2 -1 0 1NMS AXIS 1: 65 %
-2
-1
0
1
NM
S A
XIS
2: 2
8 %
B
G
A
A
G
B
A
A
G
B
B
G
A
A
A
G
G
A
A
G
G
A
B
G
A
B
G
G
G
G
B
G
G
A
A
AA
G
A
G
G
A
B
A
G
G
G
A
A
FIG. 66. Overlay of cation-exchange capacity on pimples in an NMS ordination.
113
DISCUSSION
Transect ordinations
Pimple and main dune transects. A surprising result of the ordinations of pimple
and reference transects is that water table position was often of secondary importance,
considering the hypothesized importance of that ‘free surface’. Perhaps this is because
there were a disproportionate number of xeric reference plots.
Other factors that were highly correlated with ordination axes, however, may
reflect the influence of water. Because the routinely flooded plots had gleyed soils and
smelled of sulfides, I assumed they were anoxic and reducing environments and predicted
that ammonium should be the most common inorganic nitrogen source there (Pearsall
1938, Brown et al. 1990, Chambers et al. 1992, Tobias et al. 2001, Brady and Weil 2002).
Indeed, marsh, shrub, and summit transects did sort themselves along axes highly
correlated with ammonium concentrations.
A likely source of many of the mineral and metal micronutrients is deposition
from the atmosphere rather than the ocean, either as particulates or in precipitation. Some
elements such as boron and potassium are often found in high concentrations in coastal
landscapes because they are both deposited from salt spray. Pimples are on the interior of
the island and sheltered from spray, except in major storms, and it seems likely that the
presence of some elements in and around pimples is a historical artifact of such events.
Many mineral nutrients are easily leached and would likely be less abundant in more
well-drained soils (Boyce 1954, Willis and Yemm 1961, Brooks and DeWall 1976,
Westman 1983, Bricker 1993, Bardgett et al. 2001).
114
Inspection of ordination axes to which water has a correlation coefficient greater
than 0.5 (absolute value) reveals that most mineral nutrients are correlated to axes in the
same direction (and often to a similar extent) as water table position. This suggests that
mineral nutrients are being leached from drier microhabitats but are less labile in wet
areas (Bricker 1993, Khedr and Lovett-Doust 2000, Bardgett et al. 2001, Shumway and
Banks 2001). NOx concentrations were usually negatively associated with water; this
could be due to inputs from M. cerifera nitrogen fixers and litter (Vitousek and Howarth
1991, Semones 1994, Wijnholds and Young 2000, Shumway and Banks 2001).
Furthermore, there may be even more indirect relationships to water availability among
other environmental factors. For example, calcium often complexes with soil organic
matter, and organic matter is typically highest in wet soils whose anoxic conditions
reduce decomposition rates (Khedr and Lovett-Doust 2000, Brady and Weil 2002).
Organic matter in the mineral horizon and organic horizon thickness, nevertheless, were
relatively unimportant to explaining diversity in most ordinations. This is surprising
considering that, (1) after soil moisture, soil organic content was an easily observable
factor differentiating microhabitat types in the field and (2) organic matter is important
for water and nutrient retention in sandy soils (Brady and Weil 2002).
Phosphorus, iron, manganese, and copper had conspicuously low contributions to
explaining differences among transects. The latter three are generally deposited on
coastal dunes in sufficient amounts for plant growth by sea water and salt spray, but
pimples and interior dunes are probably too far inland to receive much spray and have a
low overwash frequency (Boyce 1954, Bricker 1993, Hayden et al. 1995). Salt spray is
usually the sole source of phosphorus in coastal dunes, and it is typically not supplied in
115
amounts optimal for plant growth (Boyce 1954). Whether or not these nutrients are
limited, they appear to be distributed too uniformly or too haphazardly to be of use in
delineating similar groupings of transects.
In all ordinations, reference transects formed groupings that were distinct from
pimple transects. The difference in environmental conditions between pimples and main
dunes could be inferred as a very important driver of the differences in assemblage
composition described in the previous chapter.
Pimple transects. The extra physiographic factors available in pimple transect
data along with salinity surpass nutrient concentrations as important descriptors of
variation between transects. It is not surprising that elevation and water table position are
important descriptors of variation among pimple transects because transects were chosen
and stratified based on elevation and water presence. The importance of salinity in
ordinating transects was surprising because, although pimples were on a coastal barrier
island, they are in a freshwater ecosystem. Most coastal dunes, moreover, have low
salinities (Boyce 1954), and it would be reasonable to assume that salinity would not vary
much among pimples.
It is noteworthy, nevertheless, that many of the same nutrients important in the
ordination of dune and pimple transects together were also important to ordination axes
for pimple transects alone. Differences in nutrients like boron, magnesium, and calcium
varied enough between pimple transects to be useful discriminators in ordinations of
them.
116
Species ordinations
The purpose for ordination of transects based on environmental factors was
simply to describe how transects varied. The one caveat to be made in interpreting
ordinations of species based on environmental factors is that it is difficult to determine
whether relationships are due to coincidence or causation (ter Braak and Barendregt
1985, ter Braak 1986). It is likely that in cases where a factor is less or more important to
species ordinations than to transect ordinations, a relationship between that factor and
species distributions is not just spatial and coincidental. The coincidental spatial
distributions of species and environmental conditions do not necessarily indicate a
causative mechanism (Bray and Curtis 1957, Gauch 1973, Gauch et al. 1977, Gauch
1982b, Minchin 1987, Clarke 1993, Økland 1996, Trejo-Torres and Ackerman 2002,
Palmer 2007). Nutrients that explained the most variation may co-occur with a particular
kind of species because of other environmental conditions and could give some species
competitive advantages.
Pimple and dune species. Depending on the test used, my a priori evaluations of
species’ affinity to water predicted groupings fairly well. That notwithstanding, it is
noteworthy that explanatory contribution to ordinations of nutrients like boron,
magnesium, and calcium and cation-exchange capacity were generally higher than water
table position, even more so than in transect ordinations.
Boron was a major predictor of species presence in most ordinations, and tended
to be associated with hydrophytes. As stated previously, high boron concentration is
found in soils influenced by ocean water (Boyce 1954, Boon and MacIntyre 1968,
Brooks and DeWall 1976, Rozema et al. 1992). Boron may be more plentiful in wet areas
117
of the island directly through sea water upwelling and storm water overwash and
indirectly as it is leached from dunes into ground water (Boon and MacIntyre 1968,
Brooks and DeWall 1976, Bardgett et al. 2001). While higher concentrations of boron in
the marsh is coincidental, it probably also determines species distributions. Although an
essential micronutrient, boron is toxic to most plants in amounts only ten times that of
optimal fertilizing concentrations (Rozema et al. 1992). Rozema et al. (1992)
demonstrated that six graminoid and forb halophyte species (including a species of
Spartina) were generally more tolerant of high levels of boron than glycophytes,
probably as an adaptation to the high concentration of boron in sea water. Although
swales between dunes on Hog Island are essentially freshwater marshes, many of the
dominant hydrophytes are salt tolerant or even facultatively halophytic, e.g. Spartina
patens and Distichlis spicata, as are some uncommon species, e.g. Typha angustifolia
(Kearney 1904, Boyce 1954, Radford et al. 1968, Shumway 1995). Species that can
survive both inundated and saline conditions, i.e. halophytic hydrophytes or vice versa,
may be at the greatest competitive advantage for life in the swale marshes of Hog Island.
Magnesium and calcium are readily supplied to dune ecosystems by salt spray,
but leaching and plant uptake may still make them limiting depending on history and
location of the dune (Kearney 1904, Olsson-Seffer 1909, Boyce 1954, Gorham 1958,
Willis and Yemm 1961, Willis 1963, Pemadasa and Lovell 1974, Hester and
Mendelssohn 1990, Bardgett et al. 2001, Shumway and Banks 2001). Older dunes that
are removed from the influence of salt spray and inundation, as in the pimple and main
dunes of this study, have been found to have much lower concentrations of both cations
(Boyce 1954, Willis and Yemm 1961). Differences in availability of magnesium and
118
calcium have been implicated in dominance shifts and growth responses in some of the
same species and ones similar to them (Clayton 1972, van der Valk 1974, Hester and
Mendelssohn 1990, Khedr and Lovett-Doust 2000). For example, fertilization of dunes
with macronutrients (N, P, K) and micronutrients (Ca and Mg ; if severely limited)
elicited a shift in dominance from a beach-colonizing grass (Ammophila sp.) to a
generalist grass with higher nutrient requirements (Festuca rubra) (Gorham 1958, Willis
1963). Magnesium and calcium-related alkalinity is important to growth of endangered
basiphilous swale species in the Netherlands; one of those species, Samolus valerandi, is
also an uncommon member of the Hog Island marsh flora (Bekker et al. 1999, Lammerts
et al. 1999, Lammerts et al. 2001).
In these results, importance of calcium and magnesium increases along the
continuum from xerophytic to hydrophytic species. This suggests a more or less
coincidental association between calcium, magnesium, and species growing in particular
zones: 1) Xeric summits should have the lowest amounts of the two cations since losses
from leaching outpace the inputs from precipitation or occasional storm-driven salt spray
in these sheltered areas; 2) hydric marshes receive the leachates; and 3) shrub zones may
retain intermediate levels of cations in the thick organic layer (Kearney 1904, Olsson-
Seffer 1909, Evans 1953, Boyce 1954, Clayton 1972, van der Valk 1974, Gorham et al.
1979, Shumway and Banks 2001, Brady and Weil 2002).
Magnesium and calcium are most likely to be important to species distribution
where they are most limited, in xeric summits. For example, the grass Ammophila
breviligulata, which is a dominant species on mainline dune summits, is rarely found on
pimple summits, whereas Festuca rubra is relatively more abundant on pimples than
119
main dunes (pers. obs.). This could be a situation similar to the one already described
(Willis 1963): In pimple summits (vs. main dune summits) there is an enhanced edge
influence of the shrub zone and its higher nutrient retention in the organic layer. Indeed,
F. rubra occurs nearly exclusively along the shrub edges in the main dunes (pers. obs.; F.
Day, unpublished data).
Concentrations of magnesium, calcium, and boron are correlated with each other
and with water table position (although this is a weaker correlation). There is therefore
the possibility that these minerals have no direct effect on plant distributions and are
simply indicating the importance of water alone. Based on the importance of these
nutrients to plant health, however, I assert that the importance of boron, calcium and
magnesium to assemblage composition explains some of the importance of water .
Zinc was of intermediate importance in explaining species distributions. This is
due partly to the aforementioned syndrome of high marsh – low summit concentrations
due to leaching. Zinc has been known to interfere with phosphorus utilization in plants
and may therefore have a causative role to play in species distributions (Hester and
Mendelssohn 1990, Bricker 1993).
The nutrients that would seem most likely to influence species distributions were
not indicated as such by the analyses. Nitrogen and phosphorus are the major limiting
nutrients in dune ecosystems (Willis and Yemm 1961, Willis 1963, Keefe and Boynton
1973, Pemadasa and Lovell 1974, Gorham et al. 1979, Westman 1983, Hester and
Mendelssohn 1990, Vitousek and Howarth 1991, Bardgett et al. 2001, Pausas and Austin
2001, Shumway and Banks 2001). It was noteworthy that 1) some xeric species (e.g.,
Ammophila breviligulata, and especially Aralia spinosa) were associated with higher
120
ammonium levels than most xerophytes and 2) ammonium concentration varied
considerable among hydrophytes (e.g., Centella erecta and Hydrocotyle verticellata; Fig.
55). Since ammonium levels should be highest in wet soils, I expected to see species
affinities for ammonium follow a water-affinity gradient, similar to boron (Fig. 50).
Although it is likely that ability to use ammonium varies among the species sampled,
there could be other reasons for the association that may or may not be causative. For
example I frequently noticed rabbit droppings in the dune with A. spinosa and C. erecta,
and this could have produced higher than normal ammonium levels for the relatively dry
soils there. Whether plants or fertilizer came first cannot be determined conclusively.
Iron, manganese, and copper are essential micronutrients that are usually not
limited in coastal systems but become toxic in high amounts or in reduced forms (Jones
and Etherington 1971, Jones 1972a, 1972b, Bricker 1993). Perhaps these elements are too
sparsely distributed to have an appreciable effect.
Pimple species. The only major differences between factors determined to be
important in pimple and main dune species ordinations and ordinations of species only is
the increased importance of water availability (water table position + elevation), cation-
exchange capacity, and potassium availability. The importance of water to species
distribution is self-evident. The importance of elevation, not only to water availability,
but also in terms of exposure, organic matter accumulation, etc. is similarly easy to see
(Olsson-Seffer 1909).
Since the soil on the dunes is a uniform, well-sorted sand, cation-exchange
capacity probably represents the contributions of the organic layer and organic matter
content to nutrient retention (Lammerts et al. 1999, Brady and Weil 2002). This is
121
probably why CEC was more important to the ordinations than either organic matter
content or O-horizon.
Of the three plant macronutrients, potassium is least likely to be limited in a
coastal system. As with other nutrients, its importance to hydric and mesic species is
partially coincidental due to leaching and organic layer retention (Gorham 1958, Willis
1963, Jones 1975a). It could have a potential role to play in iron or manganese toxicity in
the marshes (Jones 1972a, 1972b, 1975a). It has been shown to influence growth in some
dune species, especially when input is limited by lack of salt spray (Boyce 1954, Gorham
1958, 1961, Willis and Yemm 1961, Willis 1963, Clayton 1972, Hester and Mendelssohn
1990).
The other physiographic factors added to the pimple species data set, slope and
aspect (divided here into eastern and northern exposure), are proxies for other factors
such as wind exposure and insolation, and have been long known to influence plant
assemblages on dunes (Olsson-Seffer 1909). The lack of importance of these factors
suggests that pimples are too protected from exposure to prevailing winds or salt spray
for them to make a difference (Boyce 1954).
One important determinant of plant assemblage structure can not be inferred from
this study: interspecific interactions. The importance of competition and facilitation
between plants in coastal ecosystems to diversity and species composition is well-
established, especially in salt-marshes (Hacker, 1999; Bertness, 1994; Bertness, 1991 }.
For example, greater coverage of Spartina patens in xeric areas of main dunes vs.
pimples may be explained by it being a poorer competitor on pimples. Perhaps some
122
factor (e.g., shading in the interior of the pimple from Myrica litter) gives other species a
competitive advantage.
CONCLUSIONS
Kruckeberg (1969) wrote that “The edaphic factor — physical and chemical
properties of soils — can elicit sharp discontinuities in plants. Sharp discontinuities
between soils of highly contrasting lithological origin exert marked selective effects on
floras.” Although organic matter, water content, or redox potential may vary in different
areas and at different depths, there is effectively only one soil on Hog Island and
Parramore Island. There are, however, distinct plant assemblages on and between the
dunes, both linear main dune and pimple. Those borders appear to be drawn along
edaphic and hydrodynamic lines more than physiographic or historical ones. I assert this
because 1) water and nutrient availability were more important to differentiating transects
and species than slope or aspect and 2) different-aged main dune plots sorted themselves
similarly to each other.
As hypothesized, fresh water availability was an important factor delineating
changes in plant assemblages. Only a few species on the island, most notably Spartina
patens, demonstrate an ability to grow well in both wet and dry areas. Because fresh
water is believed to be a driving force behind most of the ecology of the islands, it was
surprising to find other factors, e.g. boron, taking such a major role in describing
variation. The indirect effect of fresh water on these other factors (e.g. nutrient leaching
or nitrogen reduction) cannot be ignored. Overall these findings suggest that nutrient
availability as influenced by water is the main cause of shifts in plant assemblage
composition, not simply water availability or nutrient availability alone.
123
CHAPTER 5
CONCLUSIONS
WATER AND OTHER DRIVERS
My first hypothesis was that fresh water availability was the main driving force
behind plant assemblage variation on the barrier islands. My results do indicate that the
influence of fresh water on plant assemblages is profound; that was really never in
question. Nevertheless, I have found evidence that suggests other factors may be
similarly important. Although this is most likely because of those factors’ interaction
with water, it is noteworthy that the influence of water on plant assemblages is not due
simply to differing water requirements of the species involved. I base these conclusions
on the implementation of my three research goals.
Goal 1: pimple plant assemblage descriptions
Plant assemblage composition was markedly different between habitat zones on
pimples. It is reasonable to conclude that these differences are because of fresh water, but
there were differences within habitat zones that were not readily explainable by water
availability. Marsh transects were not uniform combinations of the same hydrophytic
graminoids. Rather, the presence of dominant species was patchy and proportions of
species varied greatly across the landscape. Summit transects with similar water
availability had wide differences in species richness and diversity. Even shrub transects
were not all monocultures of Morella cerifera. Similar species, such as Centella and
Hydrocotyle, demonstrated apparent differences in habitat preferences, although both
were hydrophytic.
124
Goal 2: environmental factors
Transects and species could be grouped by both fresh water availability as well as
nutrients, such as boron and potassium. The explanatory value of some of these variables
(particularly potassium) could either result from historical events or be driven by outliers.
It is impossible to prove that any patterns between species and nutrient distributions are
either the result of coincidence or underlying mechanisms. The regularity of the
association between some factors (e.g. boron concentration) and species or groups of
species suggests that there are some significant causative relationships besides water
relations driving changes in assemblage structure.
Goal 3: comparing main dunes to pimples
There were differences in diversity measures and assemblage structure between
pimple and dune transects. Those differences were most noticeable between main dune
and pimple summits. This suggests an influence of fresh water availability since the water
table was generally closer to the surface of pimple summits than main dune summits.
Differences between main dune and pimple transects were also explained by other
environmental factors besides just water, e.g. boron and cation-exchange capacity. I
suggest this is largely due to the influence of Morella cerifera. The ratio of shrub thicket
perimeter to open summit area is much higher in pimples. For pimple summit
assemblages, this could lead to 1) more nutrient input from relatively nitrogen-rich leaf
litter, both directly from decomposition, and indirectly through increased nutrient
retention; 2) more mesic conditions due to shading; and 3) differences in drainage due to
M. cerifera’s root system.
125
SPATIAL RELATIONSHIPS
My second hypothesis, that pimple size and location would be determinants of
diversity had no support from my results. There were other spatial patterns of assemblage
structure, however. Besides the obvious differences in diversity and species composition
between transects within pimples, there were differences among pimples, between
pimples and main dunes, and between Hog Island and Parramore Island Pimples.
STATISTICS
Although most of my study took place in a roughly 0.5 × 1 km section of a barrier
island, describing plant assemblage dynamics there rapidly grew complex. The ordination
techniques I used (especially CCA and NMS) were successful in simplifying the patterns
of diversity and environmental factors I encountered on the dunes. Although there was
still a net of interactions, affinities, and associations, these ordinations helped to untie
many of the knots in it. In the search for significant P values, it can be easy to lose sight
of the fact that the goal of using statistics in ecology should be to objectively detect and
describe real patterns. The techniques I used here were neither new nor particularly rare,
but they still are not used as much as they perhaps should, considering their ease of use
and interpretation.
CONCLUSION
Although there may be barrier islands elsewhere in the world with small mounded
dunes, pimples as they exist in Virginia’s climate and flora are unique. I was first drawn
to study them because of their uniqueness and because I wanted to know how
assemblages could be packed so tightly into such a small area. I found that fresh water is
not the only important ecological factor and that pimple assemblages are not as similar to
126
main dune assemblages as they may first appear. Although pimples are different from the
main dunes, I recommend they be used as experimental units.
At other LTER sites, studying the interactions of abiotic resources and plant
species in ‘insular’ communities has improved understanding and management of
imperiled ecosystems. Tundra tree islands in Niwot Ridge are important sinks for soil
nutrients; flow of ground water and nutrients between tree islands in the Everglades is a
crucial factor in those communities (Holtmeier and Broll 1992, Troxler Gann et al. 2005).
Studying the geological and biological processes that make pimples different from the
main dunes would be a considerable contribution to the knowledge of the VCR and other
coastal barrier ecosystems, especially with respect to succession, species interactions, and
dune formation. Such knowledge could also be important for management of narrowly
endemic and patchy communities.
Furthermore, continuing research could unravel the question of how pimples
form. Past hypotheses were mostly based on geological processes; one proposed
biological process, animal excavation, is not likely, based on my experience (Rich 1934,
Melton 1935, Dietz 1945, Cross 1964). Two facts indicate that pimple formation is
initiated by geological events and completed through biological succession. Their
position inside the main dunes, away from the ocean, suggests that they are fragments of
foredunes whose development has been arrested by the subsequent formation of another
line of dunes. The interaction between the abiotic conditions on pimples and the plants
that colonize them are likely responsible for the physiographic and phytosociological
differences between pimples and the main dunes. A good course for future work on
127
pimples would be to examine the role of mycorrhizal associations in their formation and
establishment (Koske and Polson 1984, Al-Agely and Reeves 1995).
128
APPENDIX
RESULTS OF ORDINATIONS NOT PRESENT IN MAIN TEXT
TABLE 19. Correlation of environmental factors with the first three axes of a Bray-Curtis ordination of main dunes and pimples (Pearson’s r).
Axis 1: 50 % Axis 2: 15 % Axis 3: 10 % NH4 0.59 CEC 0.7 P 0.435 water table -0.516 Ca 0.679 Fe -0.348 Zn -0.364 B 0.616 CaSat 0.326 B -0.327 Mg 0.614 NH4 -0.291 Mg -0.309 NH4 0.56 MgSat -0.276 NOx 0.307 Zn 0.431 Zn 0.269 MgSat -0.288 MgSat -0.428 NOx -0.262 Cu -0.284 Mn 0.413 OM -0.231 KSat 0.285 NOx 0.403 water table 0.218 P -0.269 K 0.363 Cu 0.207 Ca -0.253 OM 0.352 Ca 0.122 CEC -0.243 KSat 0.269 B 0.12 Mn 0.228 P 0.255 KSat -0.077 Fe -0.193 Fe -0.215 CEC 0.06 K 0.099 CaSat 0.154 Mn 0.05 OM -0.06 Cu 0.15 K -0.046 CaSat 0.01 water table 0.094 Mg -0.002
TABLE 20. Pearson’s r correlation coefficients for the first three axes of a principal components analysis of main dune and pimple transects versus environmental factors.
Axis 1: 31 % Axis 2: 19 % Axis 3: 18 % Mg -0.965 P 0.743 MgSat -0.92 CEC -0.952 K -0.665 Fe -0.788 Ca -0.886 KSat -0.611 Mn 0.677 B -0.852 Cu 0.597 CaSat 0.668 OM -0.723 NOx -0.575 water table -0.561 Zn -0.655 Zn 0.533 NOx 0.265 water table -0.551 CaSat 0.51 Ca 0.256 NOx -0.512 OM -0.505 NH4 0.237 K -0.358 water table 0.347 KSat 0.21 NH4 -0.347 B 0.321 P 0.163 Cu -0.289 Fe -0.18 CEC 0.166 Fe -0.274 CEC -0.136 K 0.156 P -0.213 NH4 -0.133 OM -0.108 Mn -0.113 Mn 0.078 Cu -0.037 CaSat 0.055 MgSat 0.05 B -0.023 MgSat -0.056 Mg -0.026 Mg -0.037 KSat -0.003 Ca 0.023 Zn -0.019
129
TABLE 21. Pearson’s r correlation coefficients for three axes of a Bray-Curtis ordination of pimple transects based on environmental factors.
Axis 1: 54 % Axis 2: 19 % Axis 3: 8 % Salinity -0.877 water table -0.577 K -0.879 water table -0.577 NOx 0.576 KSat -0.846 Elevation 0.513 elevation 0.556 CEC -0.554 CaSat 0.443 NH4 0.462 MgSat 0.358 MgSat -0.402 o horizon -0.455 Fe 0.324 o horizon -0.383 CEC -0.439 o horizon -0.294 Mn 0.38 Cu -0.384 Mg -0.277 Fe -0.376 Zn -0.351 CaSat 0.276 Mg -0.319 salinity -0.349 B -0.265 CEC -0.254 Mg -0.349 Ca -0.241 Zn -0.251 K -0.336 water table -0.19 B -0.235 Ca -0.327 salinity 0.183 K -0.181 B -0.289 elevation 0.177 Slope 0.162 slope 0.267 P -0.129 OM -0.16 KSat -0.201 Cu -0.098 NOx 0.137 P -0.185 NOx 0.082 KSat -0.107 CaSat 0.181 east 0.08 NH4 0.101 Mn 0.159 NH4 -0.072 Ca -0.094 east -0.089 north 0.054 Cu -0.053 OM -0.071 Mn -0.031 North 0.05 MgSat -0.04 Zn -0.017 P -0.044 Fe -0.041 slope -0.012 TABLE 22. Pearson’s r correlation coefficients for three axes of a principal components analysis ordination of pimple transects based on environmental factors.
Axis 1: 29 % Axis 2: 18 % Axis 3: 11 % Ca 0.917 CaSat 0.866 KSat 0.897 B 0.912 MgSat -0.806 K 0.863 CEC 0.85 OM -0.778 MgSat 0.193 water table 0.82 Fe -0.771 Fe 0.114 Mg 0.779 Mn 0.547 CEC 0.096 Zn 0.738 salinity -0.514 CaSat 0.075 Elevation -0.732 P 0.484 Zn 0.044 P 0.724 Mg -0.416 salinity 0.042 Cu 0.659 o horizon -0.399 slope 0.028 o horizon 0.654 water table -0.337 NOx 0.02 NOx -0.428 elevation 0.287 Cu 0.014 Slope -0.386 K -0.271 NH4 0.013 CaSat 0.227 CEC -0.249 Mn 0.01 MgSat -0.22 slope -0.206 o horizon 0.008 Fe -0.205 Cu 0.181 water table 0.008 OM -0.193 KSat -0.182 elevation 0.005 Salinity 0.178 Zn 0.165 P 0.002 NH4 0.176 NH4 -0.11 B 0.002 K 0.129 B -0.103 OM 0.001 Mn 0.106 NOx -0.107 Ca 0.001 North -0.089 Ca 0.071 north 0.001 East -0.027 north 0.051 Mg 0
130
TABLE 23. Pearson’s r correlation coefficients between environmental factors and three axes of a Bray-Curtis ordination of pimple and dune species.
Axis 1: 34 % Axis 2: 40 % Axis 3: 18 % r r r CEC 0.838 NOx -0.765 Mg 0.562 B 0.829 KSat -0.751 Ca 0.541 Mg 0.82 K -0.67 OM 0.502 Zn 0.775 MgSat 0.543 KSat -0.498 Ca 0.744 Mg 0.529 B 0.495 OM 0.645 B 0.516 CEC 0.473 P 0.495 Ca 0.487 NH4 0.442 Fe 0.484 Zn 0.483 K -0.429 Cu 0.341 CaSat 0.481 NOx -0.383 water table 0.322 Fe 0.445 MgSat 0.361 K 0.225 Cu 0.441 Zn 0.336 NOx -0.157 NH4 -0.435 CaSat 0.318 Mn -0.151 CEC 0.374 Fe 0.274 CaSat -0.145 OM 0.372 water table 0.191 MgSat 0.131 P 0.36 P 0.144 NH4 -0.096 water table 0.263 Cu 0.07 KSat 0.02 Mn -0.24 Mn -0.044
TABLE 24. Pearson’s r correlation coefficients between environmental factors and three axes of a principal components analysis of pimple and dune species.
Axis 1: 39 % Axis 2: 20 % Axis 3: 13 % r r r B 0.935 K -0.88 Mn -0.661 Mg 0.922 KSat -0.837 CaSat -0.65 Zn 0.843 NOx -0.813 Cu 0.438 CEC 0.822 CaSat 0.608 Ca -0.437 Ca 0.816 MgSat 0.526 OM -0.391 OM 0.698 NH4 -0.462 P 0.392 Fe 0.695 CEC -0.446 Fe 0.372 P 0.558 Mn -0.31 K 0.336 MgSat 0.517 OM -0.295 NH4 -0.334 NOx -0.508 Ca -0.263 MgSat 0.315 Cu 0.505 Mg -0.242 CEC -0.298 KSat -0.45 Zn -0.225 KSat 0.28 water table 0.433 B -0.217 Mg -0.233 Mn -0.309 water table -0.139 water table 0.167 K -0.26 Cu 0.12 Zn 0.108 CaSat 0.115 P 0.024 B -0.074 NH4 0.002 Fe 0.018 NOx 0.033
131
TABLE 25. Pearson’s r correlation coefficients between environmental factors and three axes of a Bray-Curtis ordination of pimple species.
Axis 1: 58 % Axis 2: 17 % Axis 3: 11 % r r R Salinity -0.923 KSat -0.924 NH4 -0.732 Mg -0.651 K -0.917 O horizon -0.518 water table -0.636 CEC -0.68 elevation 0.372 Elevation 0.537 MgSat 0.476 Mn 0.358 CEC -0.505 CaSat 0.433 water table -0.33 B -0.499 O horizon -0.404 slope -0.248 Ca -0.487 Mg -0.374 NOx -0.22 Zn -0.474 B -0.287 salinity 0.211 Fe -0.402 Ca -0.284 Fe 0.187 NOx 0.363 Fe 0.272 OM 0.18 East -0.348 Elevation 0.251 Zn 0.176 P -0.321 water table -0.237 MgSat -0.149 Mn 0.293 NOx 0.206 north -0.147 O horizon -0.285 Zn 0.189 P 0.145 north 0.266 OM 0.186 Cu 0.122 Cu -0.265 Slope -0.17 east 0.101 OM -0.256 NH4 0.136 CEC 0.083 K -0.145 P -0.113 CaSat 0.083 CaSat 0.095 North 0.101 Mg 0.084 slope 0.084 Salinity 0.082 Ca 0.081 KSat -0.069 East -0.038 KSat 0.065 MgSat -0.03 Mn 0.005 B 0.066 NH4 -0.021 Cu 0.007 K 0.046
132
TABLE 26. Pearson’s r correlation coefficients between environmental factors and three axes of a principal components analysis ordination of pimple species.
Axis 1: 35 % Axis 2: 17 % Axis 3: 14 % r r R Ca -0.962 Fe -0.802 K 0.88 B -0.944 OM -0.779 KSat 0.868 CEC -0.886 MgSat -0.768 CaSat -0.625 Mg -0.883 Mn 0.747 Zn -0.603 water table -0.859 CaSat 0.703 slope 0.475 P -0.815 North 0.475 Cu -0.447 elevation 0.779 Salinity -0.334 CEC 0.41 Zn -0.67 Elevation 0.32 O horizon 0.4 salinity -0.667 NOx -0.294 P -0.335 Cu -0.658 water table -0.282 MgSat -0.225 O horizon -0.576 East 0.28 salinity -0.15 NOx 0.525 Mg -0.269 Fe -0.132 MgSat 0.5 P 0.196 Mg 0.118 K -0.349 Slope 0.174 B -0.11 east -0.307 Ca 0.159 north -0.09 KSat -0.289 Zn -0.12 Ca -0.086 OM 0.249 O horizon -0.123 NOx -0.075 CaSat -0.212 NH4 0.097 OM -0.048 NH4 -0.201 B -0.075 east 0.039 north 0.139 Cu -0.068 NH4 -0.043 slope 0.138 KSat 0.054 elevation -0.027 Fe 0.11 CEC -0.026 Mn 0.016 Mn 0.101 K 0.011 water table -0.016
133
LITERATURE CITED
Adler, P. B., C. M. D'Antonio, and J. T. Tunison. 1998. Understory succession following
a dieback of Myrica faya in Hawai'i Volcanoes National Park. Pacific Science.
52:69-78.
Aguiar, M. R., and O. E. Sala. 1999. Patch structure, dynamics and implications for the
functioning of arid ecosystems. Trends in Ecology & Evolution 14:273-277.
Al-Agely, A. K., and F. B. Reeves. 1995. Inland sand dune mycorrhizae: effects of soil
depth, moisture, and pH on colonization of Oryzopsis hymenoides. Mycologia
87:54.
Alatalo, R., and R. Alatalo. 1977. Components of diversity: multivariate analysis with
interaction. Ecology 58:900-906.
Allen, R. B., R. K. Peet, and W. L. Baker. 1991. Gradient analysis of latitudinal variation
in southern Rocky Mountain forests. Journal of Biogeography 18:123-139.
Anthonsen, K. L., L. B. Clemmensen, and J. H. Jensen. 1996. Evolution of a dune from
crescentic to parabolic form in response to short-term climatic changes: Rabjerg
Mile, Skagen Odde, Denmark. Geomorphology 17:63-77.
Avis, A. M., and R. A. Lubke. 1996. Dynamics and succession of coastal dune vegetation
in the Eastern Cape, South Africa. Landscape and Urban Planning 34:237-253.
Badger, C. J., and R. Kellam. 1989. The Barrier Islands: A Photographic History of Life
on Hog, Cobb, Smith, Cedar, Parramore, Metompkin, & Assateague. Stackpole
Books, Harrisburg, PA.
134
Bailey, N., R. C. Kochel, and C. R. Carlson. 1998. Barrier Island Landform and
Vegetation Response to Coastal Process Variables on the Virginia Coast Reserve.
Geological Society of America Abstracts with Programs 30:2.
Bannister, P. 1966. The use of subjective estimates of cover-abundance as the basis for
ordination. The Journal of Ecology 54:665.
Bardgett, R. D., J. M. Anderson, V. Behan-Pelletier, L. Brussaard, D. C. Coleman, C.
Ettema, A. Moldenke, J. P. Schimel, and D. H. Wall. 2001. The influence of soil
biodiversity on hydrological pathways and the transfer of materials between
terrestrial and aquatic ecosystems. Ecosystems 4:421.
Bate, G., and M. Ferguson. 1996. Blowouts in coastal foredunes. Landscape and Urban
Planning 34:215-224.
Bazzaz, F. A. 1996. Plants in Changing Environments, Linking physiological, population,
and community ecology. Cambridge University Press, Cambridge.
Bekker, R. M., E. J. Lammerts, A. Schutter, and A. P. Grootjans. 1999. Vegetation
development in dune slacks: the role of persistent seed banks. Journal of
Vegetation Science 10:745-754.
Bellows, A. S., J. F. Pagels, and J. C. Mitchell. 2001. Macrohabitat and microhabitat
affinities of small mammals in a fragmented landscape on the upper coastal plain
of Virginia. American Midland Naturalist 146:345-360.
Belsky, A. J., and R. G. Amundson. 1986. Sixty years of successional history behind a
moving sand dune near Olduvai Gorge, Tanzania. Biotropica 18:231-235.
Benedict, J. B. 1984. Rates of tree-island migration, Colorado Rocky Mountains, USA.
Ecology 65:820-823.
135
Bertness, M. D. 1991a. Interspecific interactions among high marsh perennials in a New
England salt marsh. Ecology 72:125 - 137.
Bertness, M. D. 1991b. Zonation of Spartina patens and Spartina alterniflora in New
England salt marsh. Ecology 72:138 - 148.
Bertness, M. D., and A. M. Ellison. 1987. Determinants of pattern in a New England salt
marsh plant community. Ecological Monographs 57:129 - 147.
Bertness, M. D., and S. D. Hacker. 1994. Physical stress and positive associations among
marsh plants. The American Naturalist 144:363.
Bertness, M. D., and G. H. Leonard. 1997. The role of positive interactions in
communities: lessons from intertidal habitats. Ecology 78:1976 - 1989.
Bertness, M. D., and S. M. Yeh. 1994. Cooperative and competitive interactions in the
recruitment of marsh elders. Ecology 75:2416 - 2429.
Bonham, C. D. 1989. Measurements for Terrestrial Vegetation. John Wiley & Sons, New
York.
Boon, J. D., III, and W. G. MacIntyre. 1968. The boron-salinity relationship in estuarine
sediments of the Rappahannock River, Virginia. Chesapeake Science 9:21.
Booth, R. E., and J. P. Grime. 2003. Effects of genetic impoverishment on plant
community diversity. Journal of Ecology 91:721-730.
Boyce, S. G. 1954. The salt spray community. Ecological Monographs 24:29.
Brady, N. C., and R. R. Weil. 2002. The Nature and Properties of Soils, 13th edition.
Prentice Hall, Peramus, NJ.
Bray, J. R., and J. T. Curtis. 1957. An ordination of the upland forest communities of
southern Wisconsin. Ecological Monographs 27:325-349.
136
Brewer, J. S., and M. D. Bertness. 1996. Disturbance and intraspecific variation in the
clonal morphology of salt marsh perennials. Oikos 77:107 - 116.
Brewer, J. S., T. Rand, J. M. Levine, and M. D. Bertness. 1998. Biomass allocation,
clonal dispersal, and competitive success in three salt marsh plants. Oikos 82:347.
Bricker, S. B. 1993. The history of Cu, Pb, and Zn inputs to Narragansett Bay, Rhode
Island as recorded by salt-marsh sediments. Estuaries 16:589.
Brinson, M. M. 1996. The role of reference wetlands in functional assessment and
mitigation. Ecological Applications 6:69-76.
Brooks, D. J., and A. E. DeWall. 1976. Boron concentration in Chesapeake Bay
sediments, paleosalinity and baymouth uplift. Chesapeake Science 17:221.
Brown, J. F. 1997. Effects of experimental burial on survival, growth, and resource
allocation of three species of dune plants. Journal of Ecology 85:151-158.
Brown, R., E. Stone, and V. Carlisle. 1990. Soils. Pages 35-69 in R. Myers and J. Ewel,
editors. Ecosystems of Florida. University Press of Florida, Gainesville, FL USA.
Burbanck, M. P., and D. L. Phillips. 1983. Evidence of plant succession on granite
outcrops of the Georgia Piedmont. American Midland Naturalist 109:94-104.
Burton, P. J., and F. A. Bazzaz. 1995. Ecophysiological responses of tree seedlings
invading different patches of old-field vegetation. Journal of Ecology 83:99-112.
Callahan, J. T. 1984. Long-term ecological research. BioScience 34:363-367.
Chambers, R. M., H. J.W., and O. W.E. 1992. Ammonium and phosphate dynamics in a
Virginia salt marsh. Estuaries 15:349-359.
Chao, A. 1987. Estimating the population size for capture-recapture data with unequal
catchability. Biometrics 43:783.
137
Chao, A. 1989. Estimating population size for sparse data in capture-recapture
experiments. Biometrics 45:427.
Chazdon, R. L., R. K. Colwell, J. S. Denslow, and M. R. Guarigata. 1998. Statistical
methods for estimating species richness of woody regeneration in primary and
secondary rain forests of northeastern Costa Rica. Pages 285-309 in F. Dallmeier
and J. A. Comiskey, editors. Forest Biodiversity Research, Monitoring and
Modeling. UNESCO, Paris.
Chiarucci, A., V. De Dominicis, and J. B. Wilson. 2001. Structure and floristic diversity
in permanent monitoring plots in forest ecosystems of Tuscany. Forest Ecology
and Management 141:201-210.
Christian, R. R., M. Naldi, and V. P. 1998. Construction and analysis of static, structured
models of nitrogen cycling in coastal ecosystems. Pages 162-195 in Mathematical
Modeling in Microbial Ecology. Chapman Hall, New York.
Clarke, K. R. 1993. Non-parametric multivariate analyses of changes in community
structure. Austral Ecology 18:117-143.
Clayton, J. L. 1972. Salt spray and mineral cycling in two California coastal ecosystems.
Ecology 53:74.
Clements, F. E. 1936. Nature and structure of the climax. Journal of Ecology 24:252-284.
Colinvaux, P. 1993. Ecology 2. John Wiley & Sons, Inc., New York.
Colwell, R. K., and J. A. Coddington. 1994. Estimating terrestrial biodiversity through
extrapolation. Philosophical Transactions of the Royal Society of London B
Biological Sciences 345:101-118.
138
Colwell, R. K., C. X. Mao, and J. Chang. 2004. Interpolating, extrapolating, and
comparing incidence-based species accumulation curves. Ecology 85:2717-2727.
Conn, C. E., and F. P. Day. 1993. Belowground biomass patterns on a coastal barrier
island in Virginia. Bulletin of the Torrey Botanical Club 120:121-127.
Conn, C. E., and F. P. Day. 1997. Root decomposition across a barrier island
chronosequence: litter quality and environmental controls. Plant and Soil
195:351-364.
Connor, E. F., and E. D. McCoy. 1979. The statistics and biology of the species-area
relationship. American Naturalist 113:791-833.
Costa, C. S. B., J. C. Marangoni, and A. M. G. Azevedo. 2003. Plant zonation in
irregularly flooded salt marshes: Relative importance of stress tolerance and
biological interactions. The Journal of Ecology 91:951 - 965.
Cowles, H. C. 1899. The ecological relations of the vegetation on a structural basis.
Ecology 32:172-229.
Crawford, E. R., and D. R. Young. 1998a. Gap dynamics within shrub thickets on an
Atlantic Coast barrier island. Ameican Midland Naturalist 140:68-77.
Crawford, E. R., and D. R. Young. 1998b. Spatial/temporal variations in shrub thicket
soil seed banks on an Atlantic Coast barrier island. American Journal of Botany
85:1739-1744.
Crist, T. O., and J. A. Veech. 2006. Additive partitioning of rarefaction curves and
species–area relationships: unifying α-, β- and γ-diversity with sample size and
habitat area. Ecology Letters 9:923-932.
139
Crist, T. O., J. A. Veech, J. C. Gering, and K. S. Summerville. 2003. Partitioning species
diversity across landscapes and regions: a hierarchical analysis of α, β, and γ
diversity. American Naturalist 162:734.
Cross, C. L. 1964. The Parramore Island mounds of Virginia. Geographical Review
54:502-515.
Curtis, J. T., and R. P. McIntosh. 1951. An upland forest continuum in the prairie-forest
border region of Wisconsin. Ecology 32:476.
Dale, M. R. T. 1999. Spatial Pattern Analysis in Plant Ecology. Cambridge University
Press, Cambridge.
Davis, M. L. 1980. Variation in form and growth rate of Engelmann spruce at tree line.
American Midland Naturalist 104:383 - 386.
Dawson, G. W. P. 1951. A method for investigating the relationship between the
distribution of individuals of different species in a plant community. Ecology
32:332.
Day, F. P. 1996. Effects of nitrogen availability on plant biomass along a barrier island
dune chronosequence. Castanea 61:369-381.
Day, F. P., E. Crawford, and J. J. Dilustro. 2001. Plant biomass change along a coastal
barrier island dune chronosequence over a six-year period. Journal of the Torrey
Botanical Society 128:197-207.
de Castro, F. 1995. Computer simulation of the dynamics of a dune system. Ecological
Modelling 78:205-217.
Diamond, J. 1988. Factors controlling species diversity: overview and synthesis. Annals
of the Missouri Botanical Garden 75:117-129.
140
Dietz, R. S. 1945. The small mounds of the Gulf coastal plain. Science 102:596-597.
Dilustro, J. J., and F. P. Day. 1997. Aboveground biomass and net primary production
along a Virginia barrier island chronosequence. Amer. Mid. Nat. 137:27-38.
Dolan, R., and B. Hayden. 1981. Storms and shoreline configuration. Journal of
Sedimentary Petrology 51:737-744.
Dolan, R., H. Lins, and B. Hayden. 1988. Mid-Atlantic coastal storms. Journal of Coastal
Research 4:417-433.
Dueser, R. D. 1990. Biota of the Virginia barrier islands: symposium introduction.
Virginia Journal of Science 41:4.
Duever, M. J., and L. A. Riopelle. 1983. Successional sequences and rates on tree Islands
in the Okefenokee Swamp. American Midland Naturalist 110:186 - 193.
Ehrenfeld, J. G. 1990. Dynamics and processes of barrier island vegetation. Reviews in
Aquatic Sciences 2:437-480.
Ehrenfeld, J. G. 1997. Invasion of deciduous forest preserves in the New York
metropolitan region by Japanese barberry (Berberis thunbergii DC.). Journal of
the Torrey Botanical Society 124:210-215.
Erickson, D. L. 1994. Salt tolerance and the potential for ocean dispersal of a barrier
island strand glycophyte, Strophostyles umbellata (Fabaceae). M.S. Thesis.
Virginia Commonwealth University, Richmond, VA.
Erickson, D. L., and D. R. Young. 1995. Salinity response, distribution, and possible
dispersal of a barrier island strand glycophyte, Strophostyles umbellata
(Fabaceae). Bulletin of the Torrey Botanical Club 122:95-100.
141
Erwin, R. M. 1996. Dependence of waterbirds and shorebirds on shallow-water habitats
in the mid-Atlantic coastal region: an ecological profile and management
recommendations. Estuaries 19:213-219.
Evans, L. T. 1953. The ecology of the halophytic vegetation at Lake Ellesmere, New
Zealand. The Journal of Ecology 41:106.
Fahrig, L. 1990. Determinants of local population size in patchy habitats. Theoret.
Population Biology. 34:194-213.
Fahrig, L., B. Hayden, and R. Dolan. 1993. Distribution of barrier island plants in relation
to overwash disturbance: a test of life history theory. Journal of Coastal Research
9:403-412.
Faith, D. P., P. R. Minchin, and L. Belbin. 1987. Compositional dissimilarity as a robust
measure of ecological distance. Vegetatio 69:57-68.
Fauth, J. E., J. Bernardo, M. Camara, W. J. Resetarits, Jr., J. V. Buskirk, and S. A.
McCollum. 1996. Simplifying the jargon of community ecology: a conceptual
approach. The American Naturalist 147:282.
Fearnehough, W., M. A. Fullen, D. J. Mitchell, I. C. Trueman, and J. Zhang. 1998.
Aeolian deposition and its effect on soil and vegetation changes on stabilised
desert dunes in northern China. Geomorphology 23:171-182.
Field, D. P. 1999. Shrub establishment on a Virginia barrier island: spatial and temporal
variations in arbuscular mycorrhizae. MS Thesis. Virginia Commonwealth
University, Richmond, VA.
Franklin, J. F., C. S. Bledsoe, and J. T. Callahan. 1990. Contributions of the Long-Term
Ecological Research Program. BioScience 40:509-523.
142
Frego, K. A., and T. J. Carleton. 1995. Microsite tolerance of four bryophytes in a mature
black spruce stand: Reciprocal transplants. Bryologist 98:452-458.
Gauch, H. G., Jr. 1973. A quantitative evaluation of the Bray-Curtis ordination. Ecology
54:829-836.
Gauch, H. G., Jr. 1982a. Multivariate analysis in community ecology. Cambridge
University Press, Cambridge.
Gauch, H. G., Jr. 1982b. Noise reduction by eigenvector ordinations. Ecology 63:1643-
1649.
Gauch, H. G., Jr., and E. L. Stone. 1979. Vegetation and soil pattern in a mesophytic
forest at Ithaca, New York. American Midland Naturalist 102:332-345.
Gauch, H. G., Jr., and R. H. Whittaker. 1972a. Coenocline Simulation. Ecology 53:446-
451.
Gauch, H. G., Jr., and R. H. Whittaker. 1972b. Comparison of ordination techniques.
Ecology 53:868-875.
Gauch, H. G., Jr., R. H. Whittaker, and T. R. Wentworth. 1977. A comparative study of
reciprocal averaging and other ordination techniques. Journal of Ecology 65:157-
174.
Gauch, H. G. J., and G. B. Chase. 1974. Fitting the Gaussian curve to ecological data.
Ecology 55:1377-1381.
Giannini, N. P. 2003. Canonical Phylogenetic Ordination. Systematic Biology 52:684-
695.
143
Glasser, J. E. 1985. Successional trends on tree islands in the Okefenokee Swamp as
determined by interspecific association analysis. American Midland Naturalist
113:287-293.
Gleason, H. A. 1925. Species and Area. Ecology 6:66-74.
Godfrey, P. J., S. P. Leatherman, and R. Zaremba. 1979. A geobotanical approach to
classification of barrier beach systems. Pages 99-126 in S. P. Leatherman, editor.
Barrier Islands: From the Gulf of St. Lawrence to the Gulf of Mexico. Academic
Press, Inc., New York, New York.
Goodall, D. W. 1954. Objective methods for the classification of vegetation. III. An essay
in the use of factor analysis. Australian Journal of Botany 2:304-324.
Gorham, E. 1958. Soluble salts in dune sands from Blakeney Point in Norfolk. The
Journal of Ecology 46:373.
Gorham, E. 1961. The chemical composition of some waters from dune slacks at
Sandscale, North Lancashire. The Journal of Ecology 49:79.
Gorham, E., P. M. Vitousek, and W. A. Reiners. 1979. The regulation of chemical
budgets over the course of terrestrial ecosystem succession. Annual Review of
Ecology and Systematics 10:53.
Gotelli, N. J. 2001. Research frontiers in null model analysis. Global Ecology and
Biogeography 10:337-343.
Hacker, S. D., and M. D. Bertness. 1995. Morphological and physiological consequences
of a positive plant interaction. Ecology 76:2165 - 2175.
Hacker, S. D., and M. D. Bertness. 1999. Experimental evidence for factors maintaining
plant species diversity in a New England salt marsh. Ecology 80:2064 - 2073.
144
Harley, C. D. G., and M. D. Bertness. 1996. Structural interdependence: An ecological
consequence of morphological responses to crowding in marsh plants. Functional
Ecology 10:654 - 661.
Hayden, B. P., R. D. Dueser, J. T. Callahan, and H. H. Shugart. 1991. Long-term research
at the Virginia Coast Reserve: Modeling a highly dynamic environment.
BioScience 41:310-318.
Hayden, B. P., and J. A. M. Hayden. 1994. The land must change to stay the same.
Virginia Explorer 1994:2-7.
Hayden, B. P., M. C. F. V. Santos, G. Shao, and R. C. Kochel. 1995. Geomorphological
controls on coastal vegetation at the Virginia Coast Reserve. Geomorphology
13:283-300.
Hester, M. W., and I. A. Mendelssohn. 1990. Effects of macronutrient and micronutrient
additions on photosynthesis, growth parameters, and leaf nutrient concentrations
of Uniola paniculata and Panicum amarum. Botanical Gazette 151:21.
Hester, M. W., I. A. Mendelssohn, and K. L. McKee. 1996. Intraspecific variation in salt
tolerance and morphology in the coastal grass Spartina patens (Poaceae).
American Journal of Botany 83:1521-1527.
Hill, M. O. 1973a. Diversity and evenness: a unifying notation and its consequences.
Ecology 54:427.
Hill, M. O. 1973b. Reciprocal averaging: an eigenvector method of ordination. The
Journal of Ecology 61:237.
Hofer, U., L.-F. Bersier, and D. Borcard. 2004. Relating niche and spatial overlap at the
community level. Oikos 106:366-376.
145
Holtmeier, F.-K., and G. Broll. 1992. The influence of tree islands and microtopography
on pedoecological conditions in the forest-alpine tundra ecotone on Niwot Ridge,
Colorado Front Range, U.S.A. Arctic and Alpine Research 24:216 - 228.
Hosier, P. E., and T. E. Eaton. 1980. The impact of vehicles on dune and grassland
vegetation on a south-eastern North Carolina barrier beach. Journal of Applied
Ecology 17:173-182.
Houle, G. 1997. Interactions between resources and abiotic conditions control plant
performance on subarctic coastal dunes. American Journal of Botany 84:1729-
1737.
Huggett, R. J. 1998. Soil chronosequences, soil development, and soil evolution: a critical
review. Catena 32:155-172.
Johnson, A. F., and M. G. Barbour. 1990. Dunes and Maritime Forests. Pages 429-480 in
R. L. Myers and J. J. Ewel, editors. Ecosystems of Florida. University of Central
Florida Press, Orlando.
Jones, R. 1972a. Comparative studies of plant growth and distribution in relation to
waterlogging: V. The uptake of iron and manganese by dune and dune slack
plants. Journal of Ecology 60:131-139.
Jones, R. 1972b. Comparative studies of plant growth and distribution in relation to
waterlogging: VI. The effect of manganese on the growth of dune and dune slack
plants. Journal of Ecology 60:141-145.
Jones, R. 1975a. Comparative studies of plant growth and distribution in relation to
waterlogging: IX. The uptake of potassium by dune and dune slack plants. Journal
of Ecology 63:859-866.
146
Jones, R. 1975b. Comparative studies of plant growth and distribution in relation to
waterlogging: V. The uptake of phosphorus by dune and dune slack plants.
Journal of Ecology 63:109-116.
Jones, R., and J. R. Etherington. 1971. Comparative studies of plant growth and
distribution in relation to waterlogging: IV. The growth of dune and dune slack
plants. Journal of Ecology 59:793-801.
Kachi, N., and T. Hirose. 1983. Limiting nutrients for plant growth in coastal sand dune
soils. Journal of Ecology 71:937-944.
Kearney, T. H. 1904. Are plants of sea beaches and dunes true halophytes? Botanical
Gazette 37:424.
Keefe, C. W., and W. R. Boynton. 1973. Standing crop of salt marshes surrounding
Chincoteague Bay, Maryland-Virginia. Chesapeake Science 14:117.
Kent, M., and J. Ballard. 1988. Trends and problems in the application of classification
and ordination methods in plant ecology. Vegetatio 78:109-124.
Kerley, G. I. H., A. McLachlan, and J. G. Castley. 1996. Diversity and dynamics of
bushpockets in the Alexandria Coastal Dunefield, South Africa. Landscape and
Urban Planning 34:255-266.
Khedr, A.-H., and J. Lovett-Doust. 2000. Determinants of floristic diversity and
vegetation composition on the islands of Lake Burollos, Egypt. Applied
Vegetation Science 3:147.
Kochel, R. C., and L. A. Wampfler. 1989. Relative role of overwash and aeolian
processes on washover fans, Assateague Island, Virginia-Maryland. Journal of
Coastal Research 5:453-475.
147
Koske, R. E., and W. R. Polson. 1984. Are VA mycorrhizae required for sand dune
stabilization? BioScience 34:420.
Kourtev, P. S., J. G. Ehrenfeld, and W. Z. Huang. 1998. Effects of exotic plant species on
soil properties in hardwood forests of New Jersey. Water Air and Soil Pollution.
July 105:493-501.
Kruckeberg, A. R. 1969. The implications of ecology for plant systematics. Taxon 18:92-
120.
Kushlan, J. A. 1990. Freshwater Marshes. in R. L. Myers and J. J. Ewel, editors.
Ecosystems of Florida. University of Central Florida Press, Orlando, FL.
Lammerts, E. J., C. Maas, and A. P. Grootjans. 2001. Groundwater variables and
vegetation in dune slacks. Ecological Engineering 17:33-47.
Lammerts, E. J., D. M. Pegtel, A. P. Grootjans, and A. v. d. Veen. 1999. Nutrient
limitation and vegetation changes in a coastal dune slack. Journal of Vegetation
Science 10:111-122.
Lande, R. 1996. Statistics and partitioning of species diversity, and similarity among
multiple communities. Oikos 76:5-13.
Lee, S.-M., and A. Chao. 1994. Estimating population size via sample coverage for
closed capture-recapture models. Biometrics 50:88.
Levy, G. F. 1990. Vegetation dynamics on the Virginia barrier islands. Virginia Journal
of Science 41:300-306.
Lichter, J. 1998. Primary succession and forest development on coastal Lake Michigan
sand dunes. Ecological Monographs 68:487-510.
148
Longino, J. T., J. Coddington, and R. K. Colwell. 2002. The ant fauna of a tropical rain
forest: estimating species richness three different ways. Ecology 83:689.
Loveless, C. M. 1959. A study of the vegetation in the Florida Everglades. Ecology 40:1-
9.
MacArthur, R., H. Recher, and M. Cody. 1966. On the relation between habitat selection
and species diversity. The American Naturalist 100:319-332.
Magurran, A. E. 1988. Ecological Diversity and Its Measurement. Princeton University
Press, Princeton, N.J.
Marr, J. W. 1977. The development and movement of tree islands near the upper limit of
tree growth in the southern Rocky Mountains. Ecology 58:1159 - 1164.
Martin, D. W., and D. R. Young. 1997. Small-scale distribution and salinity response of
Juniperus virginiana on an Atlantic Coast barrier island. Canadian Journal of
Botany 75:77-85.
McCaffrey, C. A., and R. D. Dueser. 1990a. Plant associations on the Virginia barrier
islands. Virginia Journal of Science 41:4.
McCaffrey, C. A., and R. D. Dueser. 1990b. Preliminary vascular flora for the Virginia
barrier islands. Virginia Journal of Science 41:4.
McCune, B., and M. J. Mefford. 1999. Multivariate Analysis of Ecological Data. PC-
ORD for Windows version 4.1. MjM Software, Gleneden Beach, OR, USA.
McIntosh, R. P. 1962. Pattern in a forest communuity. Ecology 43:25.
McLachlan, A., G. Kerley, and C. Rickard. 1996. Ecology and energetics of slacks in the
Alexandria coastal dunefield. Landscape and Urban Planning 34:267-276.
Melton, F. A. 1935. Vegetation and soil mounds. Geographical Review 25:431-433.
149
Meredith, D. H. 1972. Subalpine cover associations of Eutamias amoenus and Eutamias
townsendii in the Washington Cascades. American Midland Naturalist 88:348 -
357.
Minchin, P. R. 1987. An evaluation of the relative robustness of techniques for ecological
ordination. Vegetatio 69.
Morin, P. J. 1999. Community Ecology, First edition. Blackwell Science, Malden,
Massachusetts.
Mullins, G. L., and S. E. Heckendorn. 2006. Laboratory Procedures: Virginia Tech Soil
Testing Laboratory. Publication # 452-881, Virginia Cooperative Extension,
Blacksburg, VA.
Muñoz-Reinoso, J. C. 2001. Vegetation changes and groundwater abstraction in SW
Doñana, Spain. Journal of Hydrology 242:197-209.
Norcross, B. L., and D. Hata. 1990. Seasonal composition of finfish in waters behind the
Virginia barrier islands. Virginia Journal of Science 41:441-461.
Økland, R. H. 1996. Are ordination and constrained ordination alternative or
complementary strategies in general ecological studies? Journal of Vegetation
Science 7:289-292.
Olson, J. C. 1958. Rates of succession and soil changes on southern Lake Michigan sand
dunes. Botanical Gazette 199:125-170.
Olson, R. J., J. M. Briggs, J. H. Porter, G. R. Mah, and S. G. Stafford. 1999. Managing
data from multiple disciplines, scales, and sites to support synthesis and modeling.
Remote Sensing Environment 70:99-107.
150
Olsson-Seffer, P. 1909. Relation of soil and vegetation on sandy sea shores. Botanical
Gazette 47:85.
Olszewski, T. D. 2004. A unified mathematical framework for the measurement of
richness and evenness within and among multiple communities. Oikos 104:377-
387.
Orlóci, L. 1978. Multivariate Analysis in Vegetation Research. Dr. W. Junk bv
Publishers, The Hague.
Palmer, M. 2007. The Ordination Web Page: Ordination Methods for Ecologists. in
www.ordination.okstate.edu. Oklahoma State University, Stillwater.
Parker, E. R., and R. L. Sanford, Jr. 1999. The effects of mobile tree Islands on soil
phosphorus concentrations and distribution in an alpine tundra ecosystem on
Niwot Ridge, Colorado Front Range, U.S.A. Arctic, Antarctic, and Alpine
Research 31:16 - 20.
Pauker, S. J., and T. R. Seastedt. 1996. Effects of mobile tree islands on soil carbon
storage in tundra ecosystems. Ecology 77:2563-2567.
Pausas, J. G., and M. P. Austin. 2001. Patterns of plant species richness in relation to
different environments: an appraisal. Journal of Vegetation Science 12:153.
Pearsall, W. H. 1938. The soil complex in relation to plant communities: I. Oxidation-
reduction potentials in soils. Journal of Ecology 26:180-193.
Peet, R. K., R. G. Knox, J. S. Case, and R. B. Allen. 1988. Putting things in order: The
advantages of detrended correspondence analysis. American Naturalist 131:924-
934.
151
Peet, R. K., and O. L. Loucks. 1977. A gradient of southern Wisconsin forests. Ecology
58:485-499.
Peltzer, D., A., S. Wilson, D., and A. Gerry, K. 1998. Competition intensity along a
productivity gradient in a low-diversity grassland. American Naturalist 15:465-
476.
Pemadasa, M. A., and P. H. Lovell. 1974. The mineral nutrition of some dune annuals.
The Journal of Ecology 62:647.
Pennings, S. C., and R. M. Callaway. 1992. Salt marsh plant zonation: The relative
importance of competition and physical factors. Ecology 73:681 - 690.
Perumal, J. V., and M. A. Maun. 1999. The role of mycorrhizal fungi in growth
enhancement of dune plants following burial in sand. Functional Ecology 13:560-
566.
Peterson, A. T., and N. A. Slade. 1998. Extrapolating inventory results into biodiversity
estimates and the importance of stopping rules. Diversity and Distributions 4:95-
105.
Pielou, E. C. 1984. The Interpretation of Ecological Data. John Wiley & Sons, New
York.
Planty Tabacchi, A. M., E. Tabacchi, R. J. Naiman, C. Deferrari, and H. DeCamps. 1996.
Invasibility of species-rich communities in riparian zones. Conservation Biology
10:598-607.
Radford, A. E., H. E. Ahles, and C. R. Bell. 1968. Manual of the Vascular Flora of the
Carolinas. University of North Carolina Press., Chapel Hill, NC USA.
152
Rich, F. J., and W. Spackman. 1979. Modern and ancient pollen sedimentation around
tree islands in the Okefenokee Swamp. Palynology 3:219 - 226.
Rich, J. L. 1934. Soil mottlings and mounds in northeastern Texas as seen from the air.
Geographical Review 24:576-583.
Roman, C. T., and K. F. Nordstrom. 1988. The effect of erosion rate on vegetation
patterns of an east coast barrier island. Estuarine, Coastal, and Shelf Science
26:233-242.
Rozema, J., J. de Bruin, and R. A. Broekman. 1992. Effect of boron on the growth and
mineral economy of some halophytes and non-halophytes. New Phytologist
121:249.
Rust, I. C., and W. K. Illenberger. 1996. Coastal dunes: sensitive or not? Landscape and
Urban Planning 34:165-169.
Sande, E., and D. R. Young. 1992. Effect of sodium chloride on the growth and
nitrogenase activity of Myrica cerifera seedlings. New Phytologist 120:345-350.
Schroeder, P. M., B. Hayden, and R. Dolan. 1979. Vegetation changes along the United
States east coast following the Great Storm of March 1962. Environmental
Management 3:331-338.
Seastedt, T. R., and G. A. Adams. 2001. Effects of mobile tree islands on alpine tundra
soils. Ecology 82:8 - 17.
Semones, S. W. 1994. A probable VAM association in the actinorhizal shrub Myrica
cerifera on a Virginia barrier island. M.S. Thesis. Virginia Commonwealth
University, Richmond, VA.
153
Semones, S. W., and D. R. Young. 1995. VAM association in the shrub Myrica cerifera
on a Virginia, USA barrier island. Mycorrhiza 5:423-429.
Shmida, A., and M. V. Wilson. 1985. Biological determinants of species diversity.
Journal of Biogeography 12:1-20.
Shumway, S. W. 1995. Physiological integration among clonal ramets during invasion of
disturbance patches in a New England Salt Marsh. Annals of Botany 76:225-233.
Shumway, S. W., and C. R. Banks. 2001. Species distributions in interdunal swale
communities: the effects of soil waterlogging. American Midland Naturalist
145:137.
Shumway, S. W., and M. D. Bertness. 1994. Patch size effects on marsh plant secondary
succession mechanisms. Ecology 75:564 - 568.
Silver, W. L., A. E. Lugo, and M. Keller. 1999. Soil oxygen availability and
biogeochemical cycling along elevational and topographic gradients in Puerto
Rico. Biogeochemistry 44:301-328.
Small, J. K. 1933. Manual of the southeastern flora: being descriptions of the seed plants
growing naturally in Florida, Alabama, Mississippi, eastern Louisiana, Tennessee,
North Carolina, South Carolina, and Georgia. University of North Carolina Press,
Chapel Hill, NC USA.
Smith, C. W., C. Lutzow Felling, and D. E. Gardner. 1995. Myrica faya: One man's meat
is another man's poison. Boletim do Museu Municipal do Funchal:699-706.
Stevenson, M. J., and F. P. Day. 1996. Fine root biomass distribution and production
along a barrier island chronosequence. American Midland Naturalist 135:205-
217.
154
Studer-Ehrensberger, K., C. Studer, and R. M. M. Crawford. 1993. Competition at
community boundaries: Mechanism of vegetation structure in a dune-slack
complex. Functional Ecology 7:156-168.
Summerville, K. S., and T. O. Crist. 2002. Effects of timber harvest on forest
Lepidoptera: community, guild, and species responses. Ecological Applications
12:820.
Summerville, K. S., and T. O. Crist. 2005. Temporal patterns of species accumulation in
a survey of Lepidoptera in a beech-maple forest. Biodiversity and Conservation
14:3393-3406.
Summerville, K. S., T. D. Wilson, J. A. Veech, and T. O. Crist. 2006. Do body size and
diet breadth affect partitioning of species diversity? A test with forest lepidoptera.
Diversity and Distributions 12:91.
Svihla, R. D. 1930. Some mosses from southern Louisiana. The Bryologist 33:29 - 30.
Svihla, R. D. 1939. Field notes on a collection of Louisiana Hepaticae. The Bryologist
42:118 - 120.
Swan, J. M. A. 1970. An examination of some ordination problems by use of simulated
vegetational data. Ecology 51:89-102.
ter Braak, C. J. F. 1986. Canonical Correspondence Analysis: a new eigenvector
technique for multivariate direct gradient analysis. Ecology 67:1167-1179.
ter Braak, C. J. F., and L. G. Barendregt. 1985. Weighted averaging of species indicator
values: Its efficiency in environmental calibration. Mathematical Biosciences
78:57-72.
155
Thomson, J. W., Jr. 1950. Some additional records of lichens from Alaska collected by
Walter J. Eyerdam. The Bryologist 53:9 - 15.
Tilman, D. 1997. Community invasibility, recruitment limitation, and grassland
biodiversity. Ecology 78:81-92.
Tobias, C. R., I. C. Anderson, E. A. Canuel, and S. A. Macko. 2001. Nitrogen cycling
through a fringing marsh-aquifer ecotone. Marine Ecology Progress Series
210:25-39.
Tolliver, K. S., D. M. Colley, and D. R. Young. 1995. Potential inhibitory effects of
Myrica cerifera on Pinus taeda.
Tolliver, K. S., D. M. Martin, and D. R. Young. 1997. Freshwater and saltwater flooding
response for woody species common to barrier islands swales. Wetlands 17:10-
18.
Trejo-Torres, J. C., and J. D. Ackerman. 2002. Composition patterns of Caribbean
limestone forests: Are parsimony, classification, and ordination analyses
congruent? Biotropica 34:502-515.
Troxler Gann, T. G., D. L. Childers, and D. N. Rondeau. 2005. Ecosystem structure,
nutrient dynamics, and hydrologic relationships in tree islands of the southern
Everglades, Florida, USA. Forest Ecology and Management 214:11-27.
van der Valk, A. G. 1974. Mineral cycling in coastal foredune plant communities in Cape
Hatteras National Seashore. Ecology 55:1349.
Veech, J. A., K. S. Summerville, T. O. Crist, and J. C. Gering. 2002. The additive
partitioning of species diversity: recent revival of an old idea. Oikos 99:3-9.
156
Verhoeven, J. T. A., W. Koerselman, and A. F. M. Meuleman. 1996. Nitrogen- or
phosphorus-limited growth in herbaceous, wet vegetation: relations with
atmospheric inputs and management regimes. Trends in Ecology & Evolution
11:494-497.
Vitousek, P. M., and R. W. Howarth. 1991. Nitrogen limitation on land and in the sea:
how can it occur? Biogeochemistry 13:87.
Walker, B. H., and C. F. Wehrhahn. 1971. Relationships between derived vegetation
gradients and measured environmental variables in Saskatchewan wetlands.
Ecology 52:85-95.
Weller, S. G. 1989. The effect of disturbance scale on sand dune colonization by
Lithospermum caroliniense. Ecology 70:1244-1251.
Westman, W. E. 1983. Island biogeography: studies on the xeric shrublands of the Inner
Channel Islands, California. Journal of Biogeography 10:97.
Whittaker, R. H. 1953. A consideration of climax theory: The climax as a population and
pattern. Ecological Monographs 23:41-78.
Whittaker, R. H. 1956. Vegetation of the Great Smoky Mountains. Ecological
Monographs 26:1-80.
Whittaker, R. H. 1960. Vegetation of the Siskiyou Mountains, Oregon and California.
Ecological Monographs 30:279-338.
Whittaker, R. J., K. J. Willis, and R. Field. 2001. Scale and species richness: towards a
general, hierarchical theory of species diversity. Journal of Biogeography 28:453-
470.
157
Whittecar, G. R., and J. S. Emry. 1992. Hydrogeology of a regressive barrier island
segment, Bodie Island, North Carolina. Pages 189-208 in C. A. Cole and K.
Turner, editors. Barrier Island Ecology of the Mid-Atlantic Coast: A Symposium.
National Park Service Technical Report NPS/SERCAHA/NRTR-93/04.
Wijnholds, A. E., and D. R. Young. 2000. Interdependence of the host plant, Myrica
cerifera, and the actinomycete, Frankia, in a coastal environment. Journal of
Coastal Research 16:139-144.
Will-Wolf, S. 1980. Structure of corticolous lichen communities before and after
exposure to emissions from a "clean" coal-fired generating station. The Bryologist
83:281.
Willis, A. J. 1963. Braunton Burrows: The effects on the vegetation of the addition of
mineral nutrients to the dune soils. The Journal of Ecology 51:353.
Willis, A. J., and E. W. Yemm. 1961. Braunton Burrows: mineral nutrient status of the
dune soils. The Journal of Ecology 49:377.
Wilson, M. V., and A. Shmida. 1984. Measuring beta diversity with presence–absence
data. Journal of Ecology 72:1055-1064.
Young, D. R. 1992. Photosynthetic characteristics and potential moisture stress for the
actinorhizal shrub, Myrica cerifera, on a Virginia barrier island. American Journal
of Botany 79:2-7.
Young, D. R., D. L. Erickson, and S. W. Semones. 1994. Salinity and the small-scale
distribution of three common shrubs on a Virginia barrier island. Canadian
Journal of Botany 72:1365-1372.
158
Young, D. R., E. Sande, and G. A. Peters. 1992. Spatial relationships of Frankia and
Myrica cerifera on a Virginia, USA barrier island. Symbiosis 12:209-229.
Young, D. R., G. Shao, and M. M. Brinson. 1995a. The impact of the October 1991
northeaster storm on barrier island shrub thickets (Myrica cerifera). Journal of
Coastal Research 11:1322-1328.
Young, D. R., G. Shao, and J. H. Porter. 1995b. Spatial and temporal growth dynamics of
barrier island shrub thickets. American Journal of Botany 82:638-645.
Yuan, T., M. A. Maun, and W. G. Hopkins. 1993. Effects of sand accretion on
photosynthesis, leaf-water potential and morphology of two dune grasses.
Functional Ecology 7:676-682.
Zonneveld, I. S. 1999. A geomorphological based banded ('tiger') vegetation pattern
related to former dune fields in Sokoto (Northern Nigeria). Catena 37:45-56.
Zou, X., C. P. Zucca, R. B. Waide, and W. H. McDowell. 1995. Long-term influence of
deforestation on tree species composition and litter dynamics of a tropical rain
forest in Puerto Rico. Forest Ecology and Management 78:147-157.
159
VITA Brett Allen McMillan grew up in the eastern Tennessee community of Miser
Station, near the Great Smoky Mountains. During that time, he was taught to identify
many forest trees by his paternal grandfather. After high school, he began his bachelors
of arts at Berea College in Berea, Kentucky as a biology major. He worked as a
laboratory teaching assistant and learned to enjoy teaching. He was taught to identify
even more forest trees by Dr. Ralph Thompson and further developed his appreciation for
plants. He graduated with a German minor in 1996, and began a masters of science in the
Department of Botany at the University of Florida, Gainesville. For his thesis, he studied
the impact of an invasive herb, Tradescantia fluminensis, on understory communities of
urban forests in and near Gainesville, and became an active member of the Florida Exotic
Plant Pest Council. He funded his education by working as a teaching assistant, and his
interest in teaching grew. He was taught to identify many Floridian trees and other plants
by his colleagues at UF, notably J. Richard Abbot. After completing his MS in 1999, he
worked at UF, first as an electron microscopist and then as laboratory manager for the
Department of Botany. In 2001, he began work on a PhD in ecological sciences under Dr.
Frank Day at the Biology Department of Old Dominion University, Norfolk, Virginia. He
worked first as a teaching assistant and then as course coordinator for a large freshman
environmental awareness course. He presented papers at several meetings held by the
Virginia Academy of Science, Association of Southeastern Biologists, Ecological Society
of America, and Botanical Society of America. He was taught to identify many grasses,
sedges, and forbs from Hog Island by Dr. Rebecca Bray. Upon completion of his PhD, he
plans to teach others to identify plants as a liberal arts professor.